analysisDir

Manage folder-backed collections of SanPy recording analyses.

Classes

analysisDir

Manage a folder-backed collection of SanPy recordings and analyses.

The file table is loaded from the folder's SanPy HDF5 database when one is present. Otherwise, the folder is scanned for supported recording files. Recordings can be loaded eagerly with autoLoad=True or on demand with get_analysis_for_file_key() or getAnalysis().

Source code in sanpy/analysisDir.py

class analysisDir:
    """Manage a folder-backed collection of SanPy recordings and analyses.

    The file table is loaded from the folder's SanPy HDF5 database when one is
    present. Otherwise, the folder is scanned for supported recording files.
    Recordings can be loaded eagerly with ``autoLoad=True`` or on demand with
    ``get_analysis_for_file_key()`` or ``getAnalysis()``.
    """

    sanpyColumns = _sanpyColumns
    # Dict of dict of column names and bookkeeping info.

    def __init__(
        self,
        path: str | None = None,
        fileLoaderDict: dict[str, dict[str, object]] | None = None,
        autoLoad: bool = False,
        folderDepth: Optional[int] = None,
        status_callback: Callable[[str], None] | None = None,
    ) -> None:
        """Load a folder catalog and optionally load its recordings.

        Args:
            path: Path to a recording or folder.
            fileLoaderDict: File-loader definitions keyed by extension. SanPy's
                registered loaders are used when omitted.
            autoLoad: Whether to load each recording while scanning a folder.
            folderDepth: Maximum folder depth to scan. Use ``1`` for only the
                selected folder or ``None`` for unlimited recursion.
            status_callback: Optional callback for construction progress.

        Note:
            Passing a recording path creates a one-file collection rooted in that
            recording's parent folder.
        """

        self._filePath = None
        if os.path.isfile(path):
            self._filePath = path
            folderPath = os.path.split(path)[0]
        else:
            folderPath = path

        logger.info(f'{path}')

        self.path: str = folderPath  # path to folder

        self._fileLoaderDict = fileLoaderDict

        self.autoLoad = autoLoad

        self.folderDepth = folderDepth
        self._pool_frame_cache: dict[str, tuple[str, pd.DataFrame]] = {}

        self._isDirty = False
        # keep track if analysis was changed and prompt on quit

        # self._poolDf = None
        """See pool_ functions"""

        # TODO: refactor, we are not using the csv parth of this, just the filename
        # name of database file created/loaded from folder path
        self.dbFile = "sanpy_recording_db.csv"  # old but still critical to keep

        self._df = self.loadHdf()
        if self._df is None:
            # did not load h5 file
            self._df = self.loadFolder(
                loadData=autoLoad,
                status_callback=status_callback,
            )
            self._updateLoadedAnalyzed()
        elif self._fileLoaderDict is not None:
            logger.info(f'sync existing df with filePath: {self._filePath}')
            self.syncDfWithPath()

        #
        self._checkColumns()
        self._normalize_file_keys()
        self._updateLoadedAnalyzed()

    def _normalize_file_keys(self) -> None:
        """Normalize and validate every recording key in the file table.

        Raises:
            ValueError: If a key is missing, outside the directory, or duplicated.
        """
        keys: list[str] = []
        available_paths: list[str] | None = None
        for row_index in self._df.index:
            value = self._df.at[row_index, "relPath"]
            if not isinstance(value, str) or not value:
                value = str(self._df.at[row_index, "File"])
            try:
                key = _normalized_file_key(self.path, value)
            except ValueError:
                if available_paths is None:
                    available_paths = self.getFileList()
                filename = pathlib.PureWindowsPath(value).name
                matches = [
                    path
                    for path in available_paths
                    if os.path.basename(path) == filename
                ]
                if len(matches) != 1:
                    raise ValueError(
                        f'Cannot uniquely resolve legacy recording path "{value}"'
                    ) from None
                key = _normalized_file_key(self.path, matches[0])
            self._df.at[row_index, "relPath"] = key
            keys.append(key)
        duplicates = sorted({key for key in keys if keys.count(key) > 1})
        if duplicates:
            raise ValueError(f"Duplicate analysis file keys: {duplicates}")

    def get_file_key(self, row_idx: int) -> str:
        """Return the stable file key for one table row.

        Args:
            row_idx: File-table row index.

        Returns:
            Normalized path relative to the analysis directory.
        """
        return str(self._df.at[row_idx, "relPath"])

    def get_row_for_file_key(self, file_key: str) -> int | None:
        """Return the table row for a normalized file key.

        Args:
            file_key: Recording identity relative to the analysis directory.

        Returns:
            Matching row index, or ``None`` when the key is unknown.
        """
        try:
            normalized = _normalized_file_key(self.path, file_key)
        except ValueError:
            return None
        rows = self._df.index[self._df["relPath"] == normalized].tolist()
        return int(rows[0]) if rows else None

    def get_analysis_for_file_key(
        self,
        file_key: str,
        allow_auto_load: bool = True,
    ) -> Optional[sanpy.bAnalysis]:
        """Resolve and optionally load an analysis by stable file key.

        Args:
            file_key: Recording identity relative to the analysis directory.
            allow_auto_load: Whether the recording may be loaded on demand.

        Returns:
            Resolved analysis, or ``None`` when the key is unknown or cannot load.
        """
        row_index = self.get_row_for_file_key(file_key)
        if row_index is None:
            return None
        return self.getAnalysis(row_index, allowAutoLoad=allow_auto_load)

    def detect_file_keys(
        self,
        file_keys: Sequence[str],
        detection: Mapping[str, object],
        *,
        should_continue: Callable[[int, int, str], bool] | None = None,
        after_file: Callable[[str], None] | None = None,
    ) -> tuple[int, tuple[str, ...]]:
        """Detect spikes across recordings identified by stable file keys.

        Each recording receives an independent detection dictionary covering
        its complete duration. Failures do not prevent later recordings from
        being processed.

        Args:
            file_keys: Recording identities relative to this analysis directory.
            detection: Detection preset and threshold values to apply.
            should_continue: Optional callback before each recording. Arguments
                are one-based position, total count, and file key. Returning
                ``False`` cancels before that recording.
            after_file: Optional callback after one recording finishes.

        Returns:
            Number of detected recordings and the keys that failed.
        """
        detected = 0
        failed: list[str] = []
        total = len(file_keys)
        for position, file_key in enumerate(file_keys, start=1):
            if should_continue is not None and not should_continue(
                position, total, file_key
            ):
                break
            row = self.get_row_for_file_key(file_key)
            analysis = self.get_analysis_for_file_key(file_key)
            if row is None or analysis is None or analysis.loadError:
                failed.append(file_key)
                continue
            try:
                detection_for_file = copy.deepcopy(dict(detection))
                detection_for_file["startSeconds"] = 0
                detection_for_file["stopSeconds"] = analysis.fileLoader.recordingDur
                analysis.spikeDetect(detection_for_file)
                self._updateLoadedAnalyzed(row)
            except Exception:
                logger.exception(f'Batch detection failed for "{file_key}"')
                failed.append(file_key)
                continue
            detected += 1
            if after_file is not None:
                after_file(file_key)
        return detected, tuple(failed)

    def set_metadata_for_file_keys(
        self,
        file_keys: Sequence[str],
        key: str,
        value: str,
        *,
        should_continue: Callable[[int, int, str], bool] | None = None,
        after_file: Callable[[str], None] | None = None,
    ) -> tuple[int, tuple[str, ...]]:
        """Set one experimental metadata value across recordings.

        Analyses are loaded when necessary and left dirty for the existing save
        workflow. This method never persists changes.

        Args:
            file_keys: Recording identities relative to this analysis directory.
            key: Canonical experimental-metadata key.
            value: New string value.
            should_continue: Optional callback before each recording. Arguments
                are one-based position, total count, and file key. Returning
                ``False`` cancels before that recording.
            after_file: Optional callback after one recording is processed.

        Returns:
            Number of changed recordings and the keys that failed.

        Raises:
            ValueError: If ``key`` is not canonical experimental metadata.
        """
        if key not in MetaData.getMetaDataDict():
            raise ValueError(f'Unknown experimental metadata key: "{key}"')
        changed = 0
        failed: list[str] = []
        total = len(file_keys)
        for position, file_key in enumerate(file_keys, start=1):
            if should_continue is not None and not should_continue(
                position, total, file_key
            ):
                break
            row = self.get_row_for_file_key(file_key)
            analysis = self.get_analysis_for_file_key(file_key)
            if row is None or analysis is None or analysis.loadError:
                failed.append(file_key)
                continue
            try:
                previous = analysis.metaData.getMetaData(key)
                analysis.metaData.setMetaData(key, value)
                self._updateLoadedAnalyzed(row)
            except Exception:
                logger.exception(f'Batch metadata update failed for "{file_key}"')
                failed.append(file_key)
                continue
            if previous != value:
                changed += 1
            if after_file is not None:
                after_file(file_key)
        return changed, tuple(failed)

    @property
    def theseFileTypes(self) -> list[str]:
        """Return the recording extensions available for folder scanning.

        Returns:
            Extensions from the caller-supplied loader definitions, or from
            SanPy's registered loaders when no definitions were supplied.
        """
        if self._fileLoaderDict is None:
            return list(sanpy.fileloaders.getFileLoaders().keys())
        return list(self._fileLoaderDict.keys())

    def findFileRow(self, filename):
        """Find the row index of a file in the file table.

        Args:
            filename: The name of the file to find.

        Returns:
            The row index of the file in the file table, or None if the file is not found.
        """
        fileIndexList = self._df.index[self._df['File'] == filename].tolist()
        if fileIndexList:
            rowIdx = fileIndexList[0]
            return rowIdx
        else:
            logger.warning(f"Did not find file {filename} in {self._df['File'].tolist()}")

    def __iter__(self):
        """Iterate over the bAnalysis objects in the file table."""
        self._iterIdx = -1
        return self

    def __next__(self):
        """Get the next bAnalysis object in the file table."""
        self._iterIdx += 1
        if self._iterIdx >= self.numFiles:
            self._iterIdx = -1  # reset to initial value
            raise StopIteration
        else:
            return self._df.loc[self._iterIdx]["_ba"]

    def __str__(self) -> str:
        """Return a concise folder summary.

        Returns:
            Summary containing the number of files and their total duration.
        """
        totalDurSec = self._df["Dur(s)"].sum()
        theStr = f"analysisDir Num Files: {len(self)} Total Dur(s): {totalDurSec}"
        return theStr

    @property
    def isDirty(self):
        """Get the dirty state of the analysisDir.

        Returns:
            True if the analysisDir is dirty, False otherwise.
        """
        return self._isDirty

    def __len__(self) -> int:
        """Return the number of files in the collection.

        Returns:
            Number of file-table rows.
        """
        return len(self._df)

    @property
    def numFiles(self) -> int:
        """Return the number of files in the collection.

        Returns:
            Number of file-table rows.
        """
        return len(self._df)

    @property
    def shape(self):
        """
        Can't just return shape of _df, columns (like 'ba') may have been added
        Number of columns is based on self.columns
        """
        # return self._df.shape
        numRows = self._df.shape[0]
        numCols = len(self.columns)
        return (numRows, numCols)

    @property
    def loc(self):
        """Mimic pandas df.loc[]"""
        return self._df.loc

    @loc.setter
    def loc_setter(self, rowIdx, colStr, value):
        self._df.loc[rowIdx, colStr] = value

    @property
    def iloc(self):
        # mimic pandas df.iloc[]
        return self._df.iloc

    @iloc.setter
    def iLoc_setter(self, rowIdx, colIdx, value):
        self._df.iloc[rowIdx, colIdx] = value
        self._isDirty = True

    @property
    def at(self):
        # mimic pandas df.at[]
        return self._df.at

    @at.setter
    def at_setter(self, rowIdx, colStr, value):
        self._df.at[rowIdx, colStr] = value
        self._isDirty = True

    @property
    def index(self):
        return self._df.index

    @property
    def columns(self):
        # return list of column names
        return list(self.sanpyColumns.keys())

    def copy(self) -> pd.DataFrame:
        """Return an independent copy of the folder's file table.

        Returns:
            Copy of the file table, including recording metadata and status.
        """
        return self._df.copy()

    def sort_values(self, Ncol, order):
        logger.info(f"sorting by column {self.columns[Ncol]} with order:{order}")
        self._df = self._df.sort_values(self.columns[Ncol], ascending=not order)
        # print(self._df)

    @property
    def columnsDict(self):
        return self.sanpyColumns

    def columnIsEditable(self, colName):
        return self.sanpyColumns[colName]["isEditable"]

    def columnIsCheckBox(self, colName):
        """All bool columns are checkbox

        TODO: problems with using type=bool and isinstance(). Kust using str 'bool'
        """
        type = self.sanpyColumns[colName]["type"]
        # isBool = isinstance(type, bool)
        isBool = type == "bool"
        # logger.info(f'{colName} {type(type)}, type:{type} {isBool}')
        return isBool

    def getDataFrame(self) -> pd.DataFrame:
        """Return the mutable file table used by this collection.

        Returns:
            Internal file table. Use ``copy()`` when mutations should not affect
            the collection.
        """
        return self._df

    def copyToClipboard(self):
        """
        TODO: Is this used or is copy to clipboard in pandas model?
        """
        if self.getDataFrame() is not None:
            self.getDataFrame().to_clipboard(sep="\t", index=False)
            logger.info("Copied to clipboard")

    def getPathFromRelPath(self, relPath):
        """Get full path to file from relPath.

        Uses analysisDir folder path.
        """
        if relPath.startswith("/"):
            relPath = relPath[1:]

        fullFilePath = os.path.join(self.path, relPath)

        return fullFilePath

    def saveHdf(self):
        """Save file table and any number of loaded and analyzed bAnalysis.

        Set file table 'uuid' column when we actually save a bAnalysis

        Important: Order matters
            (1) Save bAnalysis first, it updates uuid in file table.
            (2) Save file table with updated uuid
        """
        start = time.time()

        df = self.getDataFrame()

        # the compressed version from the last save
        hdfFile = os.path.splitext(self.dbFile)[0] + ".h5"
        hdfFilePath = pathlib.Path(self.path) / hdfFile

        logger.info(f"Saving h5 db (will be compressed) {hdfFilePath}")

        # save each bAnalysis
        for row in range(len(df)):
            ba = df.at[row, "_ba"]
            if ba is not None:
                didSave = ba._saveHdf_pytables(hdfFilePath)
                if didSave:
                    # we are now saved into h5 file, remember uuid to load
                    df.at[row, "uuid"] = ba.uuid
                    self._pool_frame_cache.pop(self.get_file_key(row), None)

        # rebuild (L, A, S) columns
        self._updateLoadedAnalyzed()

        #
        # save file database
        logger.info(f"    saving file db with {len(df)} rows")
        # print(df)

        dbKey = os.path.splitext(self.dbFile)[0]
        df = df.drop("_ba", axis=1)  # don't ever save _ba, use it for runtime

        # hdfStore[dbKey] = df  # save it
        df.to_hdf(hdfFilePath, key=dbKey)
        with pd.HDFStore(hdfFilePath) as hdfStore:
            hdfStore.get_storer(dbKey).attrs.sanpy_format_version = SANPY_H5_FORMAT_VERSION

        #
        self._isDirty = False  # if true, prompt to save on quit

        # rebuild the file to remove old changes and reduce size
        # self._rebuildHdf()
        sanpy.h5Util._repackHdf(hdfFilePath)

        stop = time.time()
        logger.info(f"    Saving took {round(stop-start,2)} seconds")

    def loadHdf(self, path=None, verbose=False):
        """Load the database key from an h5 file.

        We do not load analysis until user clicks on row, see loadOneAnalysis()
        """
        if path is None:
            path = self.path
        self.path = path

        df = None
        hdfFile = os.path.splitext(self.dbFile)[0] + ".h5"
        hdfPath = pathlib.Path(self.path) / hdfFile
        if not hdfPath.is_file():
            return

        # logger.info(f"Loading existing folder h5 file {hdfPath}")
        # sanpy.h5Util.listKeys(hdfPath)

        _start = time.time()
        dbKey = os.path.splitext(self.dbFile)[0]

        try:
            df = pd.read_hdf(hdfPath, dbKey)
        except KeyError as e:
            # file is corrupt !!!
            logger.error(f'    Load h5 failed, did not find dbKey:"{dbKey}" {e}')

        if df is not None:
            # _ba is for runtime, assign after loading from either (abf or h5)
            df["_ba"] = None

            # fix bug during dev of ba metadata
            # df['Sex'] = 'unknown'

            if verbose:
                logger.info("    loaded db df")

            _stop = time.time()
            # logger.info(f"Loading took {round(_stop-_start,2)} seconds")

            # if we are one file then make sure file is in df

        return df

    def loadOneAnalysis(
        self,
        path: str,
        uuid: Optional[str] = None,
        allowAutoLoad: bool = True,
        verbose: bool = False,
    ) -> Optional[sanpy.bAnalysis]:
        """Load one analysis from its recording and optional HDF5 data.

        If from h5, we still need to reload sweeps !!!
        They are binary and fast, saving to h5 (in this case) is slow.

        Args:
            path: Recording path.
            uuid: Optional analysis identifier in the folder HDF5 file.
            allowAutoLoad: Load directly from the recording when HDF5 data is
                unavailable.
            verbose: Emit detailed loading messages.

        Returns:
            A valid loaded analysis, or ``None`` when the recording cannot be
            loaded.
        """
        if verbose:
            logger.info(f'path:"{path}" uuid:"{uuid}" allowAutoLoad:"{allowAutoLoad}"')

        hdfPath = self._getHdfFile()

        ba = None
        if uuid is not None and uuid:
            # load from h5
            if verbose:
                logger.info(f"    Retreiving uuid from hdf file {uuid}")

            # load from abf
            # ba = sanpy.bAnalysis(path, fileLoaderDict=self._fileLoaderDict, verbose=verbose)
            ba = sanpy.bAnalysis(path, loadData=allowAutoLoad, fileLoaderDict=self._fileLoaderDict, verbose=verbose)

            if ba.loadError:
                logger.error('Unable to load recording "%s"', path)
                return None

            # load analysis from h5 file, will fail if uuid is not in file
            ba._loadHdf_pytables(hdfPath, uuid)

        if allowAutoLoad and ba is None:
            # load from path
            ba = sanpy.bAnalysis(path, fileLoaderDict=self._fileLoaderDict, verbose=verbose)
            if verbose:
                logger.info(f"    Loaded ba from path {path} and now ba:{ba}")

        if ba is not None and ba.loadError:
            logger.error('Unable to load recording "%s"', path)
            return None
        #
        return ba

    def _getHdfFile(self):
        hdfFile = os.path.splitext(self.dbFile)[0] + ".h5"
        hdfPath = os.path.join(self.path, hdfFile)
        return hdfPath

    def _deleteFromHdf(self, uuid):
        """Delete uuid from h5 file.

        Each bAnalysis detection get a unique uuid.
        """
        if uuid is None or not uuid:
            return
        logger.info(f"deleting from h5 file uuid:{uuid}")

        _hdfFile = os.path.splitext(self.dbFile)[0] + ".h5"
        hdfPath = pathlib.Path(self.path) / _hdfFile

        # tmpHdfPath = self._getTmpHdfFile()

        removed = False
        with pd.HDFStore(hdfPath) as hdfStore:
            try:
                hdfStore.remove(uuid)
                removed = True
            except KeyError:
                logger.error(f"Did not find uuid {uuid} in h5 file {hdfPath}")

        #
        if removed:
            # will rebuild on next save
            # self._rebuildHdf()
            self._updateLoadedAnalyzed()
            self._isDirty = True  # if true, prompt to save on quit

    def loadFolder(
        self,
        path: str | None = None,
        loadData: bool = False,
        status_callback: Callable[[str], None] | None = None,
    ) -> pd.DataFrame:
        """Parse a folder and build its recording table.

        Args:
            path: Folder to scan. The configured directory is used when omitted.
            loadData: Whether to retain each loaded analysis in memory.
            status_callback: Optional callback for per-file progress messages.

        Returns:
            Dataframe containing one row for each supported recording.
        """
        logger.info("Loading folder from scratch (no h5 file)")

        start = time.time()
        if path is None:
            path = self.path
        self.path = path

        df = pd.DataFrame(columns=self.sanpyColumns.keys())
        df = self._setColumnType(df)

        # get list of all abf/csv/tif files
        fileList = self.getFileList(path)
        _numFilesToLoad = len(fileList)
        start = time.time()
        # build new db dataframe
        listOfDict = []
        for rowIdx, fullFilePath in enumerate(fileList):
            status_message = (
                f'Loading file {rowIdx+1} of {_numFilesToLoad} "{_normalized_file_key(self.path, fullFilePath)}"'
            )
            if status_callback is None:
                logger.info(status_message)
            else:
                status_callback(status_message)

            # rowDict is what we are showing in the file table
            # abb debug vue, set loadData=True
            # loads bAnalysis
            ba, rowDict = self.getFileRow(fullFilePath, loadData=loadData)

            if rowDict is None:
                logger.warning(f'error loading file {fullFilePath}')
                continue

            # as we parse the folder, don't load ALL files (will run out of memory)
            if loadData:
                rowDict["_ba"] = ba
            else:
                rowDict["_ba"] = None  # ba

            # do not assign uuid until bAnalysis is saved in h5 file
            # rowDict['uuid'] = ''

            # logger.info(f'    row:{rowIdx} relPath:{relPath} fullFilePath:{fullFilePath}')

            listOfDict.append(rowDict)

        stop = time.time()
        logger.info(f"Loading {len(listOfDict)} files took {round(stop-start,3)} seconds.")

        df = pd.DataFrame(listOfDict)
        df = self._setColumnType(df)

        return df

    def _checkColumns(self, verbose = False):
        """Check columns in loaded vs sanpyColumns (and vica versa).
        """
        if self._df is None:
            return

        # verbose = True
        loadedColumns = self._df.columns
        for col in loadedColumns:
            if col not in self.sanpyColumns.keys():
                # loaded has unexpected column, leave it
                if verbose:
                    logger.info(
                        f'did not find loaded col: "{col}" in sanpyColumns.keys() ... ignore it'
                    )
        for col in self.sanpyColumns.keys():
            if col not in loadedColumns:
                # loaded is missing expected, add it
                if verbose:
                    logger.info(
                        f'did not find sanpyColumns.keys() col: "{col}" in loadedColumns ... adding col'
                    )
                self._df[col] = ""

    def _updateLoadedAnalyzed(self, theRowIdx=None):
        """Refresh Loaded (L) and Analyzed (A) columns.

        Arguments:
            theRowIdx (int): Update just one row

        TODO: For kymograph, add rows (left, top, right, bottom) and update
        """
        if self._df is None:
            return
        for rowIdx in range(len(self._df)):
            if theRowIdx is not None and theRowIdx != rowIdx:
                continue

            ba = self._df.loc[rowIdx, "_ba"]  # Can be None

            # uuid = self._df.at[rowIdx, 'uuid']
            #
            # loaded
            if self.isLoaded(rowIdx):
                theChar = "\u2022"  # FILLED BULLET
            # elif uuid:
            #    #theChar = '\u25CB'  # open circle
            #    theChar = '\u25e6'  # white bullet
            else:
                theChar = ""
            # self._df.iloc[rowIdx, loadedCol] = theChar
            self._df.loc[rowIdx, "L"] = theChar
            #
            # analyzed
            if self.isAnalyzed(rowIdx):
                theChar = "\u2022"  # FILLED BULLET
                self._df.loc[rowIdx, "N"] = ba.numSpikes
                _numErrors = ba.numErrors
                if _numErrors is None:
                    _numErrors = ''
                # logger.warning(f'setting E to _numErrors {_numErrors}')
                self._df.loc[rowIdx, "E"] = _numErrors
            # elif uuid:
            #    #theChar = '\u25CB'
            #    theChar = '\u25e6'  # white bullet
            else:
                theChar = ""
                self._df.loc[rowIdx, "A"] = ""
            # self._df.iloc[rowIdx, analyzedCol] = theChar
            self._df.loc[rowIdx, "A"] = theChar
            #
            # saved
            if self.isSaved(rowIdx):
                theChar = "\u2022"  # FILLED BULLET
            else:
                theChar = ""
            # self._df.iloc[rowIdx, savedCol] = theChar
            self._df.loc[rowIdx, "S"] = theChar
            #
            # start(s) and stop(s) from ba detectionDict
            if self.isAnalyzed(rowIdx):
                # set table to values we just detected with

                # relPath should usually be filled in ???
                """
                relPath = self.getPathFromRelPath(ba._path)
                self._df.loc[rowIdx, 'relPath'] = relPath
                """

                # logger.info('maybe put back in')
                # print(f'    self._df.loc[rowIdx, "relPath"] is "{self._df.loc[rowIdx, "relPath"]}"')

            # aug 2023, update meta data columns
            if ba is not None:
                for k,v in ba.metaData.items():
                    self._df.loc[rowIdx, k] = v

            # TODO: remove start of ba._path that corresponds to our current folder path
            # will allow our save db to be modular
            # self._df.loc[rowIdx, 'path'] = ba._path

    def isLoaded(self, rowIdx):
        isLoaded = self._df.loc[rowIdx, "_ba"] is not None
        return isLoaded

    def isAnalyzed(self, rowIdx):
        isAnalyzed = False
        # abb 20261012 type hinting
        ba : sanpy.bAnalysis = self._df.loc[rowIdx, "_ba"]
        if ba is not None:
            try:
                isAnalyzed = ba.isAnalyzed()
            except(AttributeError) as e:
                logger.error(f'rowIdx {rowIdx} ba is "{ba}" but expecting bAnalysis_')
                logger.error('self._df:')
                print(self._df)
                return False
        return isAnalyzed

    def analysisIsDirty(self, rowIdx):
        """Analysis is dirty when there has been detection but not saved to h5."""
        isDirty = False
        ba = self._df.loc[rowIdx, "_ba"]
        if isinstance(ba, sanpy.bAnalysis):
            isDirty = ba.isDirty()
        return isDirty

    def hasDirty(self):
        """Return true if any bAnalysis in list has been analyzed but not saved (e.g. is dirty).
        """
        haveDirty = False
        numRows = len(self._df)
        for rowIdx in range(numRows):
            if self.analysisIsDirty(rowIdx):
                haveDirty = True

        return haveDirty

    def isSaved(self, rowIdx):
        uuid = self._df.at[rowIdx, "uuid"]
        return len(uuid) > 0

    def getAnalysis(self, rowIdx, allowAutoLoad=True, verbose=False) -> sanpy.bAnalysis:
        """Get bAnalysis object, will load if necc.

        Args:
            rowIdx (int): Row index from table, corresponds to row in self._df
            allowAutoLoad (bool)
        Return:
            bAnalysis
        """
        file = self._df.loc[rowIdx, "File"]
        ba = self._df.loc[rowIdx, "_ba"]
        uuid = self._df.loc[rowIdx, "uuid"]  # if we have a uuid bAnalysis is saved in h5f

        if ba is None or ba == "":
            relPath = self._df.loc[rowIdx, "relPath"]
            filePath = self.getPathFromRelPath(relPath)

            ba = self.loadOneAnalysis(
                filePath, uuid, allowAutoLoad=allowAutoLoad, verbose=verbose
            )
            # load
            if ba is None:
                logger.warning(
                    f'Did not load row {rowIdx} path: "{filePath}". Analysis was probably not saved'
                )
            else:
                self._df.at[rowIdx, "_ba"] = ba
                # does not get a uuid until save into h5
                if uuid:
                    # there was an original uuid (in table), means we are saved into h5
                    self._df.at[rowIdx, "uuid"] = uuid
                    if uuid != ba.uuid:
                        logger.error(
                            "Loaded uuid does not match existing in file table"
                        )
                        logger.error(f"  Loaded {ba.uuid}")
                        logger.error(f"  Existing {uuid}")

                #
                # update stats of table load/analyzed columns
                self._updateLoadedAnalyzed()

        return ba

    def _setColumnType(self, df):
        """Needs to be called every time a df is created.
        Ensures proper type of columns following sanpyColumns[key]['type']
        """
        # print('columns are:', df.columns)
        for col in df.columns:
            # when loading from csv, 'col' may not be in sanpyColumns
            if col not in self.sanpyColumns:
                # logger.warning(f'Column "{col}" is not in sanpyColumns -->> ignoring')
                continue
            colType = self.sanpyColumns[col]["type"]
            if colType == str:
                df[col] = df[col].replace(np.nan, "", regex=True)
                df[col] = df[col].astype(str)
            elif colType == int:
                pass
            elif colType == float:
                # error if ''
                df[col] = df[col].astype(float)
            elif colType == bool:
                df[col] = df[col].astype(bool)
            else:
                logger.warning(f'Did not parse col "{col}" with type "{colType}"')
        #
        return df

    def getFileRow(self, path, loadData=False):
        """Get dict representing one file (row in table). Loads bAnalysis to get headers.

        On load error of proper file type (abf, csv), ba.loadError==True

        Args:
            path (Str): Full path to file.
            #rowIdx (int): Optional row index to assign in column 'Idx'

        Return:
            (tuple): tuple containing:

            - ba (bAnalysis): [sanpy.bAnalysis](/api/bAnalysis).
            - rowDict (dict): On success, otherwise None.
                    fails when path does not lead to valid bAnalysis file.
        """
        if not os.path.isfile(path):
            logger.warning(f'Did not find file "{path}"')
            return None, None

        fileType = os.path.splitext(path)[1]

        if fileType not in self.theseFileTypes:  
            logger.warning(f'Did not load file type "{fileType}"')
            return None, None

        ba = sanpy.bAnalysis(path,
                             loadData=loadData,
                             fileLoaderDict=self._fileLoaderDict)

        if ba.loadError:
            logger.error(f'Error loading bAnalysis file "{path}"')
            # return None, None

        # not sufficient to default everything to empty str ''
        # sanpyColumns can only have type in ('float', 'str')
        rowDict = dict.fromkeys(self.sanpyColumns.keys(), "")
        for k in rowDict.keys():
            if self.sanpyColumns[k]["type"] == str:
                rowDict[k] = ""
            elif self.sanpyColumns[k]["type"] == float:
                rowDict[k] = np.nan

        if ba.loadError:
            return None, None

        rowDict["File"] = ba.fileLoader.filename  # os.path.split(ba.path)[1]
        rowDict["Dur(s)"] = ba.fileLoader.recordingDur
        rowDict["Channels"] = ba.fileLoader.numChannels  # Theanne
        rowDict["Sweeps"] = ba.fileLoader.numSweeps

        # here, we do not get an epoch table until the file is loaded !!!
        rowDict["Epochs"] = ba.fileLoader.numEpochs  # Theanne, data has to be loaded

        rowDict["kHz"] = ba.fileLoader.recordingFrequency
        rowDict["Mode"] = ba.fileLoader.recordingMode.value

        # add parent1, parent2, parent3
        _path, _file = os.path.split(path)
        _path, _parent1 = os.path.split(_path)
        _path, _parent2 = os.path.split(_path)
        _path, _parent3 = os.path.split(_path)
        rowDict['parent1'] = _parent1
        rowDict['parent2'] = _parent2
        rowDict['parent3'] = _parent3

        # aug 2023,  adding bAnalysis metadata columns
        for k,v in ba.metaData.items():
            rowDict[k] = v

        # remove the path to the folder we have loaded
        rowDict["relPath"] = _normalized_file_key(self.path, path)

        return ba, rowDict

    def getFileList(self,
                    path: str = None,
                    ) -> List[str]:
        """Get file paths from path.

        Uses self.theseFileTypes

        """

        if self._filePath is not None:
            logger.info(f'returning one file {self._filePath}')
            return [self._filePath]

        if path is None:
            path = self.path

        fileList = getFileList(path, self.theseFileTypes, self.folderDepth)
        return fileList

    def getRowDict(self, rowIdx):
        """
        Return a dict with selected row as dict (includes detection parameters).

        Important to return a copy as our '_ba' is a pointer to bAnalysis.

        Returns:
            theRet (dict): Be sure to make a deep copy of ['_ba'] if neccessary.
        """
        theRet = {}
        # use columns in main sanpyColumns, not in df
        # for colStr in self.columns:
        for colStr in self._df.columns:
            # theRet[colStr] = self._df.loc[rowIdx, colStr]
            theRet[colStr] = self._df.loc[rowIdx, colStr]
        # theRet['_ba'] = theRet['_ba'].copy()
        return theRet

    def appendRow(self, rowDict=None, ba=None):
        """Append an empty row."""

        rowSeries = pd.Series()
        if rowDict is not None:
            # rowSeries = pd.Series(rowDict)
            rowSeries = pd.DataFrame([rowDict])

            # self._data.iloc[row] = rowSeries
            # self._data = self._data.reset_index(drop=True)

        newRowIdx = len(self._df)  # append this row

        df = self._df
        df = pd.concat([df, rowSeries], axis=0, ignore_index=True)

        # df = pd.concat([df,rowSeries], ignore_index=True, axis=1)
        df = df.reset_index(drop=True)

        if ba is not None:
            df.loc[newRowIdx, "_ba"] = ba

        #
        self._df = df

    def unloadRow(self, rowIdx):
        self._pool_frame_cache.pop(self.get_file_key(rowIdx), None)
        self._df.loc[rowIdx, "_ba"] = None
        self._updateLoadedAnalyzed()

    def removeRowFromDatabase(self, rowIdx):
        self._pool_frame_cache.pop(self.get_file_key(rowIdx), None)
        # delete from h5 file
        uuid = self._df.at[rowIdx, "uuid"]
        self._deleteFromHdf(uuid)

        # clear uuid
        self._df.at[rowIdx, "uuid"] = ""

        self._updateLoadedAnalyzed()

    def deleteRow(self, rowIdx):
        df = self._df
        self._pool_frame_cache.pop(self.get_file_key(rowIdx), None)

        # delete from h5 file
        uuid = df.at[rowIdx, "uuid"]
        self._deleteFromHdf(uuid)

        # delete from df/model
        df = df.drop([rowIdx])
        df = df.reset_index(drop=True)
        self._df = df

        self._updateLoadedAnalyzed()

    def syncDfWithPath(self):
        """Sync path with existing df. Used to detect new/removed files.

        If we currently have just one file (self._filePath) we will trash it and load a folder

        Notes
        -----
        20231230, trying to use this to open a one file window with an exiting h5 file.
        """
        self._pool_frame_cache.clear()

        pathFileList = self.getFileList()

        # our currently loaded files
        dfFileList = self._df["File"].tolist()

        # logger.info(f'dfFileList: {dfFileList}')
        # print('    === pathFileList (on drive):')
        # print('    ', pathFileList)
        # print('    === dfFileList (in table):')
        # print('    ', dfFileList)

        addedToDf = False

        # look for files in path not in df
        for pathFile in pathFileList:
            fileName = os.path.split(pathFile)[1]
            if fileName not in dfFileList:
                logger.info(f'   Found file in path "{fileName}" not in df')

                # load bAnalysis and get df column values
                addedToDf = True

                ba, rowDict = self.getFileRow(pathFile)  # loads bAnalysis

                if rowDict is not None:
                    # listOfDict.append(rowDict)

                    # TODO: get this into getFileROw()
                    # logger.warning("bug 20220718, not sure we need this ???")
                    # print(rowDict)

                    # rowDict['relPath'] = pathFile
                    rowDict["_ba"] = None

                    self.appendRow(rowDict=rowDict, ba=None)

        # look for files in df not in path
        # for dfFile in dfFileList:
        #     if not dfFile in pathFileList:
        #         logger.info(f'Found file in df "{dfFile}" not in path')

        if addedToDf:
            df = self._df
            df = df.sort_values(
                by=["File"], axis="index", ascending=True, inplace=False
            )
            df = df.reset_index(drop=True)
            self._df = df

        self._updateLoadedAnalyzed()

    def spike_dataframe_for_row(self, row_idx: int) -> pd.DataFrame | None:
        """Return pooled spike results for one file-table row.

        A loaded analysis is always authoritative. For an unloaded row, a
        matching cached frame is reused or the saved spike results and file
        metadata are read from the folder HDF5 file without opening the
        recording.

        Args:
            row_idx: File-table row index.

        Returns:
            Copied spike results with pooled identity columns, or ``None`` when
            the row has no saved analyzed spikes.

        Raises:
            ValueError: If analyzed results have no ``spikeNumber`` column.
        """
        file_key = self.get_file_key(row_idx)
        uuid = str(self._df.at[row_idx, "uuid"])
        ba = self._df.at[row_idx, "_ba"]
        cache_result = ba is None or ba == ""

        if not cache_result:
            self._pool_frame_cache.pop(file_key, None)
            if not ba.isAnalyzed():
                return None
            one_df = ba.asDataFrame(regenerateAnalysisDataFrame=True)
        else:
            cached = self._pool_frame_cache.get(file_key)
            if cached is not None:
                if cached[0] == uuid:
                    cached_frame = cached[1]
                    return None if cached_frame.empty else cached_frame.copy()
                self._pool_frame_cache.pop(file_key, None)
            if not uuid:
                return None
            with pd.HDFStore(self._getHdfFile(), mode="r") as store:
                analysis_key = f"/{uuid}/analysisList"
                if analysis_key not in store:
                    return None
                spikes = sanpy.bAnalysisResults.analysisResultList()
                spikes.setFromListDict(store[analysis_key].to_dict("records"))
                metadata = MetaData()
                metadata_key = f"/{uuid}/metaDataDict"
                if metadata_key in store:
                    saved = store[metadata_key].to_dict("records")[0]
                    metadata.fromDict(
                        {k: v for k, v in saved.items() if k in metadata},
                        triggerDirty=False,
                    )
            one_df = sanpy.bAnalysis_.analysis_dataframe(
                spikes.asDataFrame(), metadata
            )

        if one_df is None or one_df.empty:
            if cache_result:
                self._pool_frame_cache[file_key] = (uuid, pd.DataFrame())
            return None
        if "spikeNumber" not in one_df.columns:
            raise ValueError(
                "SanPy spike results are missing the spikeNumber column"
            )

        one_df = one_df.copy()
        one_df["file_number"] = int(row_idx)
        one_df["file_key"] = file_key
        one_df["row_id"] = (
            one_df["file_number"].astype(int).astype(str)
            + ":"
            + one_df["spikeNumber"].astype(int).astype(str)
        )
        if cache_result:
            self._pool_frame_cache[file_key] = (uuid, one_df.copy())
        return one_df

    def pool_spike_dataframe(
        self,
        status_callback: Callable[[str], None] | None = None,
    ) -> pd.DataFrame | None:
        """Concatenate spike results for every analyzed file.

        Reuses per-file cached results or loads a saved analysis without raw
        sweep arrays when that file is not already in memory.

        Args:
            status_callback: Optional callback for per-file progress messages.

        Returns:
            The concatenated spike table, or ``None`` when no file has spikes.

        Raises:
            ValueError: If an analyzed file has no ``spikeNumber`` column.

        """
        frames: list[pd.DataFrame] = []
        row_indices = list(self._df.index)
        total = len(row_indices)
        for position, row_idx in enumerate(row_indices, start=1):
            file_key = self.get_file_key(int(row_idx))
            status_message = (
                f'Pooling file {position} of {total} "{file_key}"'
            )
            if status_callback is None:
                logger.info(status_message)
            else:
                status_callback(status_message)
            one_df = self.spike_dataframe_for_row(int(row_idx))
            if one_df is not None:
                frames.append(one_df)
        if not frames:
            return None
        return pd.concat(frames, ignore_index=True)

    def api_getFileHeaders(self):
        headerList = []
        df = self.getDataFrame()
        for row in range(len(df)):
            # ba = self.getAnalysis(row)  # do not call this, it will load
            ba = df.at[row, "_ba"]
            if ba is not None:
                headerDict = ba.api_getHeader()
                headerList.append(headerDict)
        #
        return headerList

Attributes

isDirty property

Get the dirty state of the analysisDir.

Returns: True if the analysisDir is dirty, False otherwise.

loc property

Mimic pandas df.loc[]

numFiles property

Return the number of files in the collection.

Returns: Number of file-table rows.

shape property

Can't just return shape of _df, columns (like 'ba') may have been added Number of columns is based on self.columns

theseFileTypes property

Return the recording extensions available for folder scanning.

Returns: Extensions from the caller-supplied loader definitions, or from SanPy's registered loaders when no definitions were supplied.

Methods:

__init__(path=None, fileLoaderDict=None, autoLoad=False, folderDepth=None, status_callback=None)

Load a folder catalog and optionally load its recordings.

Args: path: Path to a recording or folder. fileLoaderDict: File-loader definitions keyed by extension. SanPy's registered loaders are used when omitted. autoLoad: Whether to load each recording while scanning a folder. folderDepth: Maximum folder depth to scan. Use 1 for only the selected folder or None for unlimited recursion. status_callback: Optional callback for construction progress.

Note: Passing a recording path creates a one-file collection rooted in that recording's parent folder.

Source code in sanpy/analysisDir.py

def __init__(
    self,
    path: str | None = None,
    fileLoaderDict: dict[str, dict[str, object]] | None = None,
    autoLoad: bool = False,
    folderDepth: Optional[int] = None,
    status_callback: Callable[[str], None] | None = None,
) -> None:
    """Load a folder catalog and optionally load its recordings.

    Args:
        path: Path to a recording or folder.
        fileLoaderDict: File-loader definitions keyed by extension. SanPy's
            registered loaders are used when omitted.
        autoLoad: Whether to load each recording while scanning a folder.
        folderDepth: Maximum folder depth to scan. Use ``1`` for only the
            selected folder or ``None`` for unlimited recursion.
        status_callback: Optional callback for construction progress.

    Note:
        Passing a recording path creates a one-file collection rooted in that
        recording's parent folder.
    """

    self._filePath = None
    if os.path.isfile(path):
        self._filePath = path
        folderPath = os.path.split(path)[0]
    else:
        folderPath = path

    logger.info(f'{path}')

    self.path: str = folderPath  # path to folder

    self._fileLoaderDict = fileLoaderDict

    self.autoLoad = autoLoad

    self.folderDepth = folderDepth
    self._pool_frame_cache: dict[str, tuple[str, pd.DataFrame]] = {}

    self._isDirty = False
    # keep track if analysis was changed and prompt on quit

    # self._poolDf = None
    """See pool_ functions"""

    # TODO: refactor, we are not using the csv parth of this, just the filename
    # name of database file created/loaded from folder path
    self.dbFile = "sanpy_recording_db.csv"  # old but still critical to keep

    self._df = self.loadHdf()
    if self._df is None:
        # did not load h5 file
        self._df = self.loadFolder(
            loadData=autoLoad,
            status_callback=status_callback,
        )
        self._updateLoadedAnalyzed()
    elif self._fileLoaderDict is not None:
        logger.info(f'sync existing df with filePath: {self._filePath}')
        self.syncDfWithPath()

    #
    self._checkColumns()
    self._normalize_file_keys()
    self._updateLoadedAnalyzed()

__iter__()

Iterate over the bAnalysis objects in the file table.

Source code in sanpy/analysisDir.py

def __iter__(self):
    """Iterate over the bAnalysis objects in the file table."""
    self._iterIdx = -1
    return self

__len__()

Return the number of files in the collection.

Returns: Number of file-table rows.

Source code in sanpy/analysisDir.py

def __len__(self) -> int:
    """Return the number of files in the collection.

    Returns:
        Number of file-table rows.
    """
    return len(self._df)

__next__()

Get the next bAnalysis object in the file table.

Source code in sanpy/analysisDir.py

def __next__(self):
    """Get the next bAnalysis object in the file table."""
    self._iterIdx += 1
    if self._iterIdx >= self.numFiles:
        self._iterIdx = -1  # reset to initial value
        raise StopIteration
    else:
        return self._df.loc[self._iterIdx]["_ba"]

__str__()

Return a concise folder summary.

Returns: Summary containing the number of files and their total duration.

Source code in sanpy/analysisDir.py

def __str__(self) -> str:
    """Return a concise folder summary.

    Returns:
        Summary containing the number of files and their total duration.
    """
    totalDurSec = self._df["Dur(s)"].sum()
    theStr = f"analysisDir Num Files: {len(self)} Total Dur(s): {totalDurSec}"
    return theStr

analysisIsDirty(rowIdx)

Analysis is dirty when there has been detection but not saved to h5.

Source code in sanpy/analysisDir.py

def analysisIsDirty(self, rowIdx):
    """Analysis is dirty when there has been detection but not saved to h5."""
    isDirty = False
    ba = self._df.loc[rowIdx, "_ba"]
    if isinstance(ba, sanpy.bAnalysis):
        isDirty = ba.isDirty()
    return isDirty

appendRow(rowDict=None, ba=None)

Append an empty row.

Source code in sanpy/analysisDir.py

def appendRow(self, rowDict=None, ba=None):
    """Append an empty row."""

    rowSeries = pd.Series()
    if rowDict is not None:
        # rowSeries = pd.Series(rowDict)
        rowSeries = pd.DataFrame([rowDict])

        # self._data.iloc[row] = rowSeries
        # self._data = self._data.reset_index(drop=True)

    newRowIdx = len(self._df)  # append this row

    df = self._df
    df = pd.concat([df, rowSeries], axis=0, ignore_index=True)

    # df = pd.concat([df,rowSeries], ignore_index=True, axis=1)
    df = df.reset_index(drop=True)

    if ba is not None:
        df.loc[newRowIdx, "_ba"] = ba

    #
    self._df = df

columnIsCheckBox(colName)

All bool columns are checkbox

TODO: problems with using type=bool and isinstance(). Kust using str 'bool'

Source code in sanpy/analysisDir.py

def columnIsCheckBox(self, colName):
    """All bool columns are checkbox

    TODO: problems with using type=bool and isinstance(). Kust using str 'bool'
    """
    type = self.sanpyColumns[colName]["type"]
    # isBool = isinstance(type, bool)
    isBool = type == "bool"
    # logger.info(f'{colName} {type(type)}, type:{type} {isBool}')
    return isBool

copy()

Return an independent copy of the folder's file table.

Returns: Copy of the file table, including recording metadata and status.

Source code in sanpy/analysisDir.py

def copy(self) -> pd.DataFrame:
    """Return an independent copy of the folder's file table.

    Returns:
        Copy of the file table, including recording metadata and status.
    """
    return self._df.copy()

copyToClipboard()

TODO: Is this used or is copy to clipboard in pandas model?

Source code in sanpy/analysisDir.py

def copyToClipboard(self):
    """
    TODO: Is this used or is copy to clipboard in pandas model?
    """
    if self.getDataFrame() is not None:
        self.getDataFrame().to_clipboard(sep="\t", index=False)
        logger.info("Copied to clipboard")

detect_file_keys(file_keys, detection, *, should_continue=None, after_file=None)

Detect spikes across recordings identified by stable file keys.

Each recording receives an independent detection dictionary covering its complete duration. Failures do not prevent later recordings from being processed.

Args: file_keys: Recording identities relative to this analysis directory. detection: Detection preset and threshold values to apply. should_continue: Optional callback before each recording. Arguments are one-based position, total count, and file key. Returning False cancels before that recording. after_file: Optional callback after one recording finishes.

Returns: Number of detected recordings and the keys that failed.

Source code in sanpy/analysisDir.py

def detect_file_keys(
    self,
    file_keys: Sequence[str],
    detection: Mapping[str, object],
    *,
    should_continue: Callable[[int, int, str], bool] | None = None,
    after_file: Callable[[str], None] | None = None,
) -> tuple[int, tuple[str, ...]]:
    """Detect spikes across recordings identified by stable file keys.

    Each recording receives an independent detection dictionary covering
    its complete duration. Failures do not prevent later recordings from
    being processed.

    Args:
        file_keys: Recording identities relative to this analysis directory.
        detection: Detection preset and threshold values to apply.
        should_continue: Optional callback before each recording. Arguments
            are one-based position, total count, and file key. Returning
            ``False`` cancels before that recording.
        after_file: Optional callback after one recording finishes.

    Returns:
        Number of detected recordings and the keys that failed.
    """
    detected = 0
    failed: list[str] = []
    total = len(file_keys)
    for position, file_key in enumerate(file_keys, start=1):
        if should_continue is not None and not should_continue(
            position, total, file_key
        ):
            break
        row = self.get_row_for_file_key(file_key)
        analysis = self.get_analysis_for_file_key(file_key)
        if row is None or analysis is None or analysis.loadError:
            failed.append(file_key)
            continue
        try:
            detection_for_file = copy.deepcopy(dict(detection))
            detection_for_file["startSeconds"] = 0
            detection_for_file["stopSeconds"] = analysis.fileLoader.recordingDur
            analysis.spikeDetect(detection_for_file)
            self._updateLoadedAnalyzed(row)
        except Exception:
            logger.exception(f'Batch detection failed for "{file_key}"')
            failed.append(file_key)
            continue
        detected += 1
        if after_file is not None:
            after_file(file_key)
    return detected, tuple(failed)

findFileRow(filename)

Find the row index of a file in the file table.

Args: filename: The name of the file to find.

Returns: The row index of the file in the file table, or None if the file is not found.

Source code in sanpy/analysisDir.py

def findFileRow(self, filename):
    """Find the row index of a file in the file table.

    Args:
        filename: The name of the file to find.

    Returns:
        The row index of the file in the file table, or None if the file is not found.
    """
    fileIndexList = self._df.index[self._df['File'] == filename].tolist()
    if fileIndexList:
        rowIdx = fileIndexList[0]
        return rowIdx
    else:
        logger.warning(f"Did not find file {filename} in {self._df['File'].tolist()}")

getAnalysis(rowIdx, allowAutoLoad=True, verbose=False)

Get bAnalysis object, will load if necc.

Args: rowIdx (int): Row index from table, corresponds to row in self._df allowAutoLoad (bool) Return: bAnalysis

Source code in sanpy/analysisDir.py

def getAnalysis(self, rowIdx, allowAutoLoad=True, verbose=False) -> sanpy.bAnalysis:
    """Get bAnalysis object, will load if necc.

    Args:
        rowIdx (int): Row index from table, corresponds to row in self._df
        allowAutoLoad (bool)
    Return:
        bAnalysis
    """
    file = self._df.loc[rowIdx, "File"]
    ba = self._df.loc[rowIdx, "_ba"]
    uuid = self._df.loc[rowIdx, "uuid"]  # if we have a uuid bAnalysis is saved in h5f

    if ba is None or ba == "":
        relPath = self._df.loc[rowIdx, "relPath"]
        filePath = self.getPathFromRelPath(relPath)

        ba = self.loadOneAnalysis(
            filePath, uuid, allowAutoLoad=allowAutoLoad, verbose=verbose
        )
        # load
        if ba is None:
            logger.warning(
                f'Did not load row {rowIdx} path: "{filePath}". Analysis was probably not saved'
            )
        else:
            self._df.at[rowIdx, "_ba"] = ba
            # does not get a uuid until save into h5
            if uuid:
                # there was an original uuid (in table), means we are saved into h5
                self._df.at[rowIdx, "uuid"] = uuid
                if uuid != ba.uuid:
                    logger.error(
                        "Loaded uuid does not match existing in file table"
                    )
                    logger.error(f"  Loaded {ba.uuid}")
                    logger.error(f"  Existing {uuid}")

            #
            # update stats of table load/analyzed columns
            self._updateLoadedAnalyzed()

    return ba

getDataFrame()

Return the mutable file table used by this collection.

Returns: Internal file table. Use copy() when mutations should not affect the collection.

Source code in sanpy/analysisDir.py

def getDataFrame(self) -> pd.DataFrame:
    """Return the mutable file table used by this collection.

    Returns:
        Internal file table. Use ``copy()`` when mutations should not affect
        the collection.
    """
    return self._df

getFileList(path=None)

Get file paths from path.

Uses self.theseFileTypes

Source code in sanpy/analysisDir.py

def getFileList(self,
                path: str = None,
                ) -> List[str]:
    """Get file paths from path.

    Uses self.theseFileTypes

    """

    if self._filePath is not None:
        logger.info(f'returning one file {self._filePath}')
        return [self._filePath]

    if path is None:
        path = self.path

    fileList = getFileList(path, self.theseFileTypes, self.folderDepth)
    return fileList

getFileRow(path, loadData=False)

Get dict representing one file (row in table). Loads bAnalysis to get headers.

On load error of proper file type (abf, csv), ba.loadError==True

Args: path (Str): Full path to file. #rowIdx (int): Optional row index to assign in column 'Idx'

Return: (tuple): tuple containing:

- ba (bAnalysis): [sanpy.bAnalysis](/api/bAnalysis).
- rowDict (dict): On success, otherwise None.
        fails when path does not lead to valid bAnalysis file.

Source code in sanpy/analysisDir.py

def getFileRow(self, path, loadData=False):
    """Get dict representing one file (row in table). Loads bAnalysis to get headers.

    On load error of proper file type (abf, csv), ba.loadError==True

    Args:
        path (Str): Full path to file.
        #rowIdx (int): Optional row index to assign in column 'Idx'

    Return:
        (tuple): tuple containing:

        - ba (bAnalysis): [sanpy.bAnalysis](/api/bAnalysis).
        - rowDict (dict): On success, otherwise None.
                fails when path does not lead to valid bAnalysis file.
    """
    if not os.path.isfile(path):
        logger.warning(f'Did not find file "{path}"')
        return None, None

    fileType = os.path.splitext(path)[1]

    if fileType not in self.theseFileTypes:  
        logger.warning(f'Did not load file type "{fileType}"')
        return None, None

    ba = sanpy.bAnalysis(path,
                         loadData=loadData,
                         fileLoaderDict=self._fileLoaderDict)

    if ba.loadError:
        logger.error(f'Error loading bAnalysis file "{path}"')
        # return None, None

    # not sufficient to default everything to empty str ''
    # sanpyColumns can only have type in ('float', 'str')
    rowDict = dict.fromkeys(self.sanpyColumns.keys(), "")
    for k in rowDict.keys():
        if self.sanpyColumns[k]["type"] == str:
            rowDict[k] = ""
        elif self.sanpyColumns[k]["type"] == float:
            rowDict[k] = np.nan

    if ba.loadError:
        return None, None

    rowDict["File"] = ba.fileLoader.filename  # os.path.split(ba.path)[1]
    rowDict["Dur(s)"] = ba.fileLoader.recordingDur
    rowDict["Channels"] = ba.fileLoader.numChannels  # Theanne
    rowDict["Sweeps"] = ba.fileLoader.numSweeps

    # here, we do not get an epoch table until the file is loaded !!!
    rowDict["Epochs"] = ba.fileLoader.numEpochs  # Theanne, data has to be loaded

    rowDict["kHz"] = ba.fileLoader.recordingFrequency
    rowDict["Mode"] = ba.fileLoader.recordingMode.value

    # add parent1, parent2, parent3
    _path, _file = os.path.split(path)
    _path, _parent1 = os.path.split(_path)
    _path, _parent2 = os.path.split(_path)
    _path, _parent3 = os.path.split(_path)
    rowDict['parent1'] = _parent1
    rowDict['parent2'] = _parent2
    rowDict['parent3'] = _parent3

    # aug 2023,  adding bAnalysis metadata columns
    for k,v in ba.metaData.items():
        rowDict[k] = v

    # remove the path to the folder we have loaded
    rowDict["relPath"] = _normalized_file_key(self.path, path)

    return ba, rowDict

getPathFromRelPath(relPath)

Get full path to file from relPath.

Uses analysisDir folder path.

Source code in sanpy/analysisDir.py

def getPathFromRelPath(self, relPath):
    """Get full path to file from relPath.

    Uses analysisDir folder path.
    """
    if relPath.startswith("/"):
        relPath = relPath[1:]

    fullFilePath = os.path.join(self.path, relPath)

    return fullFilePath

getRowDict(rowIdx)

Return a dict with selected row as dict (includes detection parameters).

Important to return a copy as our '_ba' is a pointer to bAnalysis.

Returns: theRet (dict): Be sure to make a deep copy of ['_ba'] if neccessary.

Source code in sanpy/analysisDir.py

def getRowDict(self, rowIdx):
    """
    Return a dict with selected row as dict (includes detection parameters).

    Important to return a copy as our '_ba' is a pointer to bAnalysis.

    Returns:
        theRet (dict): Be sure to make a deep copy of ['_ba'] if neccessary.
    """
    theRet = {}
    # use columns in main sanpyColumns, not in df
    # for colStr in self.columns:
    for colStr in self._df.columns:
        # theRet[colStr] = self._df.loc[rowIdx, colStr]
        theRet[colStr] = self._df.loc[rowIdx, colStr]
    # theRet['_ba'] = theRet['_ba'].copy()
    return theRet

get_analysis_for_file_key(file_key, allow_auto_load=True)

Resolve and optionally load an analysis by stable file key.

Args: file_key: Recording identity relative to the analysis directory. allow_auto_load: Whether the recording may be loaded on demand.

Returns: Resolved analysis, or None when the key is unknown or cannot load.

Source code in sanpy/analysisDir.py

def get_analysis_for_file_key(
    self,
    file_key: str,
    allow_auto_load: bool = True,
) -> Optional[sanpy.bAnalysis]:
    """Resolve and optionally load an analysis by stable file key.

    Args:
        file_key: Recording identity relative to the analysis directory.
        allow_auto_load: Whether the recording may be loaded on demand.

    Returns:
        Resolved analysis, or ``None`` when the key is unknown or cannot load.
    """
    row_index = self.get_row_for_file_key(file_key)
    if row_index is None:
        return None
    return self.getAnalysis(row_index, allowAutoLoad=allow_auto_load)

get_file_key(row_idx)

Return the stable file key for one table row.

Args: row_idx: File-table row index.

Returns: Normalized path relative to the analysis directory.

Source code in sanpy/analysisDir.py

def get_file_key(self, row_idx: int) -> str:
    """Return the stable file key for one table row.

    Args:
        row_idx: File-table row index.

    Returns:
        Normalized path relative to the analysis directory.
    """
    return str(self._df.at[row_idx, "relPath"])

get_row_for_file_key(file_key)

Return the table row for a normalized file key.

Args: file_key: Recording identity relative to the analysis directory.

Returns: Matching row index, or None when the key is unknown.

Source code in sanpy/analysisDir.py

def get_row_for_file_key(self, file_key: str) -> int | None:
    """Return the table row for a normalized file key.

    Args:
        file_key: Recording identity relative to the analysis directory.

    Returns:
        Matching row index, or ``None`` when the key is unknown.
    """
    try:
        normalized = _normalized_file_key(self.path, file_key)
    except ValueError:
        return None
    rows = self._df.index[self._df["relPath"] == normalized].tolist()
    return int(rows[0]) if rows else None

hasDirty()

Return true if any bAnalysis in list has been analyzed but not saved (e.g. is dirty).

Source code in sanpy/analysisDir.py

def hasDirty(self):
    """Return true if any bAnalysis in list has been analyzed but not saved (e.g. is dirty).
    """
    haveDirty = False
    numRows = len(self._df)
    for rowIdx in range(numRows):
        if self.analysisIsDirty(rowIdx):
            haveDirty = True

    return haveDirty

loadFolder(path=None, loadData=False, status_callback=None)

Parse a folder and build its recording table.

Args: path: Folder to scan. The configured directory is used when omitted. loadData: Whether to retain each loaded analysis in memory. status_callback: Optional callback for per-file progress messages.

Returns: Dataframe containing one row for each supported recording.

Source code in sanpy/analysisDir.py

def loadFolder(
    self,
    path: str | None = None,
    loadData: bool = False,
    status_callback: Callable[[str], None] | None = None,
) -> pd.DataFrame:
    """Parse a folder and build its recording table.

    Args:
        path: Folder to scan. The configured directory is used when omitted.
        loadData: Whether to retain each loaded analysis in memory.
        status_callback: Optional callback for per-file progress messages.

    Returns:
        Dataframe containing one row for each supported recording.
    """
    logger.info("Loading folder from scratch (no h5 file)")

    start = time.time()
    if path is None:
        path = self.path
    self.path = path

    df = pd.DataFrame(columns=self.sanpyColumns.keys())
    df = self._setColumnType(df)

    # get list of all abf/csv/tif files
    fileList = self.getFileList(path)
    _numFilesToLoad = len(fileList)
    start = time.time()
    # build new db dataframe
    listOfDict = []
    for rowIdx, fullFilePath in enumerate(fileList):
        status_message = (
            f'Loading file {rowIdx+1} of {_numFilesToLoad} "{_normalized_file_key(self.path, fullFilePath)}"'
        )
        if status_callback is None:
            logger.info(status_message)
        else:
            status_callback(status_message)

        # rowDict is what we are showing in the file table
        # abb debug vue, set loadData=True
        # loads bAnalysis
        ba, rowDict = self.getFileRow(fullFilePath, loadData=loadData)

        if rowDict is None:
            logger.warning(f'error loading file {fullFilePath}')
            continue

        # as we parse the folder, don't load ALL files (will run out of memory)
        if loadData:
            rowDict["_ba"] = ba
        else:
            rowDict["_ba"] = None  # ba

        # do not assign uuid until bAnalysis is saved in h5 file
        # rowDict['uuid'] = ''

        # logger.info(f'    row:{rowIdx} relPath:{relPath} fullFilePath:{fullFilePath}')

        listOfDict.append(rowDict)

    stop = time.time()
    logger.info(f"Loading {len(listOfDict)} files took {round(stop-start,3)} seconds.")

    df = pd.DataFrame(listOfDict)
    df = self._setColumnType(df)

    return df

loadHdf(path=None, verbose=False)

Load the database key from an h5 file.

We do not load analysis until user clicks on row, see loadOneAnalysis()

Source code in sanpy/analysisDir.py

def loadHdf(self, path=None, verbose=False):
    """Load the database key from an h5 file.

    We do not load analysis until user clicks on row, see loadOneAnalysis()
    """
    if path is None:
        path = self.path
    self.path = path

    df = None
    hdfFile = os.path.splitext(self.dbFile)[0] + ".h5"
    hdfPath = pathlib.Path(self.path) / hdfFile
    if not hdfPath.is_file():
        return

    # logger.info(f"Loading existing folder h5 file {hdfPath}")
    # sanpy.h5Util.listKeys(hdfPath)

    _start = time.time()
    dbKey = os.path.splitext(self.dbFile)[0]

    try:
        df = pd.read_hdf(hdfPath, dbKey)
    except KeyError as e:
        # file is corrupt !!!
        logger.error(f'    Load h5 failed, did not find dbKey:"{dbKey}" {e}')

    if df is not None:
        # _ba is for runtime, assign after loading from either (abf or h5)
        df["_ba"] = None

        # fix bug during dev of ba metadata
        # df['Sex'] = 'unknown'

        if verbose:
            logger.info("    loaded db df")

        _stop = time.time()
        # logger.info(f"Loading took {round(_stop-_start,2)} seconds")

        # if we are one file then make sure file is in df

    return df

loadOneAnalysis(path, uuid=None, allowAutoLoad=True, verbose=False)

Load one analysis from its recording and optional HDF5 data.

If from h5, we still need to reload sweeps !!! They are binary and fast, saving to h5 (in this case) is slow.

Args: path: Recording path. uuid: Optional analysis identifier in the folder HDF5 file. allowAutoLoad: Load directly from the recording when HDF5 data is unavailable. verbose: Emit detailed loading messages.

Returns: A valid loaded analysis, or None when the recording cannot be loaded.

Source code in sanpy/analysisDir.py

def loadOneAnalysis(
    self,
    path: str,
    uuid: Optional[str] = None,
    allowAutoLoad: bool = True,
    verbose: bool = False,
) -> Optional[sanpy.bAnalysis]:
    """Load one analysis from its recording and optional HDF5 data.

    If from h5, we still need to reload sweeps !!!
    They are binary and fast, saving to h5 (in this case) is slow.

    Args:
        path: Recording path.
        uuid: Optional analysis identifier in the folder HDF5 file.
        allowAutoLoad: Load directly from the recording when HDF5 data is
            unavailable.
        verbose: Emit detailed loading messages.

    Returns:
        A valid loaded analysis, or ``None`` when the recording cannot be
        loaded.
    """
    if verbose:
        logger.info(f'path:"{path}" uuid:"{uuid}" allowAutoLoad:"{allowAutoLoad}"')

    hdfPath = self._getHdfFile()

    ba = None
    if uuid is not None and uuid:
        # load from h5
        if verbose:
            logger.info(f"    Retreiving uuid from hdf file {uuid}")

        # load from abf
        # ba = sanpy.bAnalysis(path, fileLoaderDict=self._fileLoaderDict, verbose=verbose)
        ba = sanpy.bAnalysis(path, loadData=allowAutoLoad, fileLoaderDict=self._fileLoaderDict, verbose=verbose)

        if ba.loadError:
            logger.error('Unable to load recording "%s"', path)
            return None

        # load analysis from h5 file, will fail if uuid is not in file
        ba._loadHdf_pytables(hdfPath, uuid)

    if allowAutoLoad and ba is None:
        # load from path
        ba = sanpy.bAnalysis(path, fileLoaderDict=self._fileLoaderDict, verbose=verbose)
        if verbose:
            logger.info(f"    Loaded ba from path {path} and now ba:{ba}")

    if ba is not None and ba.loadError:
        logger.error('Unable to load recording "%s"', path)
        return None
    #
    return ba

pool_spike_dataframe(status_callback=None)

Concatenate spike results for every analyzed file.

Reuses per-file cached results or loads a saved analysis without raw sweep arrays when that file is not already in memory.

Args: status_callback: Optional callback for per-file progress messages.

Returns: The concatenated spike table, or None when no file has spikes.

Raises: ValueError: If an analyzed file has no spikeNumber column.

Source code in sanpy/analysisDir.py

def pool_spike_dataframe(
    self,
    status_callback: Callable[[str], None] | None = None,
) -> pd.DataFrame | None:
    """Concatenate spike results for every analyzed file.

    Reuses per-file cached results or loads a saved analysis without raw
    sweep arrays when that file is not already in memory.

    Args:
        status_callback: Optional callback for per-file progress messages.

    Returns:
        The concatenated spike table, or ``None`` when no file has spikes.

    Raises:
        ValueError: If an analyzed file has no ``spikeNumber`` column.

    """
    frames: list[pd.DataFrame] = []
    row_indices = list(self._df.index)
    total = len(row_indices)
    for position, row_idx in enumerate(row_indices, start=1):
        file_key = self.get_file_key(int(row_idx))
        status_message = (
            f'Pooling file {position} of {total} "{file_key}"'
        )
        if status_callback is None:
            logger.info(status_message)
        else:
            status_callback(status_message)
        one_df = self.spike_dataframe_for_row(int(row_idx))
        if one_df is not None:
            frames.append(one_df)
    if not frames:
        return None
    return pd.concat(frames, ignore_index=True)

saveHdf()

Save file table and any number of loaded and analyzed bAnalysis.

Set file table 'uuid' column when we actually save a bAnalysis

Important: Order matters (1) Save bAnalysis first, it updates uuid in file table. (2) Save file table with updated uuid

Source code in sanpy/analysisDir.py

def saveHdf(self):
    """Save file table and any number of loaded and analyzed bAnalysis.

    Set file table 'uuid' column when we actually save a bAnalysis

    Important: Order matters
        (1) Save bAnalysis first, it updates uuid in file table.
        (2) Save file table with updated uuid
    """
    start = time.time()

    df = self.getDataFrame()

    # the compressed version from the last save
    hdfFile = os.path.splitext(self.dbFile)[0] + ".h5"
    hdfFilePath = pathlib.Path(self.path) / hdfFile

    logger.info(f"Saving h5 db (will be compressed) {hdfFilePath}")

    # save each bAnalysis
    for row in range(len(df)):
        ba = df.at[row, "_ba"]
        if ba is not None:
            didSave = ba._saveHdf_pytables(hdfFilePath)
            if didSave:
                # we are now saved into h5 file, remember uuid to load
                df.at[row, "uuid"] = ba.uuid
                self._pool_frame_cache.pop(self.get_file_key(row), None)

    # rebuild (L, A, S) columns
    self._updateLoadedAnalyzed()

    #
    # save file database
    logger.info(f"    saving file db with {len(df)} rows")
    # print(df)

    dbKey = os.path.splitext(self.dbFile)[0]
    df = df.drop("_ba", axis=1)  # don't ever save _ba, use it for runtime

    # hdfStore[dbKey] = df  # save it
    df.to_hdf(hdfFilePath, key=dbKey)
    with pd.HDFStore(hdfFilePath) as hdfStore:
        hdfStore.get_storer(dbKey).attrs.sanpy_format_version = SANPY_H5_FORMAT_VERSION

    #
    self._isDirty = False  # if true, prompt to save on quit

    # rebuild the file to remove old changes and reduce size
    # self._rebuildHdf()
    sanpy.h5Util._repackHdf(hdfFilePath)

    stop = time.time()
    logger.info(f"    Saving took {round(stop-start,2)} seconds")

set_metadata_for_file_keys(file_keys, key, value, *, should_continue=None, after_file=None)

Set one experimental metadata value across recordings.

Analyses are loaded when necessary and left dirty for the existing save workflow. This method never persists changes.

Args: file_keys: Recording identities relative to this analysis directory. key: Canonical experimental-metadata key. value: New string value. should_continue: Optional callback before each recording. Arguments are one-based position, total count, and file key. Returning False cancels before that recording. after_file: Optional callback after one recording is processed.

Returns: Number of changed recordings and the keys that failed.

Raises: ValueError: If key is not canonical experimental metadata.

Source code in sanpy/analysisDir.py

def set_metadata_for_file_keys(
    self,
    file_keys: Sequence[str],
    key: str,
    value: str,
    *,
    should_continue: Callable[[int, int, str], bool] | None = None,
    after_file: Callable[[str], None] | None = None,
) -> tuple[int, tuple[str, ...]]:
    """Set one experimental metadata value across recordings.

    Analyses are loaded when necessary and left dirty for the existing save
    workflow. This method never persists changes.

    Args:
        file_keys: Recording identities relative to this analysis directory.
        key: Canonical experimental-metadata key.
        value: New string value.
        should_continue: Optional callback before each recording. Arguments
            are one-based position, total count, and file key. Returning
            ``False`` cancels before that recording.
        after_file: Optional callback after one recording is processed.

    Returns:
        Number of changed recordings and the keys that failed.

    Raises:
        ValueError: If ``key`` is not canonical experimental metadata.
    """
    if key not in MetaData.getMetaDataDict():
        raise ValueError(f'Unknown experimental metadata key: "{key}"')
    changed = 0
    failed: list[str] = []
    total = len(file_keys)
    for position, file_key in enumerate(file_keys, start=1):
        if should_continue is not None and not should_continue(
            position, total, file_key
        ):
            break
        row = self.get_row_for_file_key(file_key)
        analysis = self.get_analysis_for_file_key(file_key)
        if row is None or analysis is None or analysis.loadError:
            failed.append(file_key)
            continue
        try:
            previous = analysis.metaData.getMetaData(key)
            analysis.metaData.setMetaData(key, value)
            self._updateLoadedAnalyzed(row)
        except Exception:
            logger.exception(f'Batch metadata update failed for "{file_key}"')
            failed.append(file_key)
            continue
        if previous != value:
            changed += 1
        if after_file is not None:
            after_file(file_key)
    return changed, tuple(failed)

spike_dataframe_for_row(row_idx)

Return pooled spike results for one file-table row.

A loaded analysis is always authoritative. For an unloaded row, a matching cached frame is reused or the saved spike results and file metadata are read from the folder HDF5 file without opening the recording.

Args: row_idx: File-table row index.

Returns: Copied spike results with pooled identity columns, or None when the row has no saved analyzed spikes.

Raises: ValueError: If analyzed results have no spikeNumber column.

Source code in sanpy/analysisDir.py

def spike_dataframe_for_row(self, row_idx: int) -> pd.DataFrame | None:
    """Return pooled spike results for one file-table row.

    A loaded analysis is always authoritative. For an unloaded row, a
    matching cached frame is reused or the saved spike results and file
    metadata are read from the folder HDF5 file without opening the
    recording.

    Args:
        row_idx: File-table row index.

    Returns:
        Copied spike results with pooled identity columns, or ``None`` when
        the row has no saved analyzed spikes.

    Raises:
        ValueError: If analyzed results have no ``spikeNumber`` column.
    """
    file_key = self.get_file_key(row_idx)
    uuid = str(self._df.at[row_idx, "uuid"])
    ba = self._df.at[row_idx, "_ba"]
    cache_result = ba is None or ba == ""

    if not cache_result:
        self._pool_frame_cache.pop(file_key, None)
        if not ba.isAnalyzed():
            return None
        one_df = ba.asDataFrame(regenerateAnalysisDataFrame=True)
    else:
        cached = self._pool_frame_cache.get(file_key)
        if cached is not None:
            if cached[0] == uuid:
                cached_frame = cached[1]
                return None if cached_frame.empty else cached_frame.copy()
            self._pool_frame_cache.pop(file_key, None)
        if not uuid:
            return None
        with pd.HDFStore(self._getHdfFile(), mode="r") as store:
            analysis_key = f"/{uuid}/analysisList"
            if analysis_key not in store:
                return None
            spikes = sanpy.bAnalysisResults.analysisResultList()
            spikes.setFromListDict(store[analysis_key].to_dict("records"))
            metadata = MetaData()
            metadata_key = f"/{uuid}/metaDataDict"
            if metadata_key in store:
                saved = store[metadata_key].to_dict("records")[0]
                metadata.fromDict(
                    {k: v for k, v in saved.items() if k in metadata},
                    triggerDirty=False,
                )
        one_df = sanpy.bAnalysis_.analysis_dataframe(
            spikes.asDataFrame(), metadata
        )

    if one_df is None or one_df.empty:
        if cache_result:
            self._pool_frame_cache[file_key] = (uuid, pd.DataFrame())
        return None
    if "spikeNumber" not in one_df.columns:
        raise ValueError(
            "SanPy spike results are missing the spikeNumber column"
        )

    one_df = one_df.copy()
    one_df["file_number"] = int(row_idx)
    one_df["file_key"] = file_key
    one_df["row_id"] = (
        one_df["file_number"].astype(int).astype(str)
        + ":"
        + one_df["spikeNumber"].astype(int).astype(str)
    )
    if cache_result:
        self._pool_frame_cache[file_key] = (uuid, one_df.copy())
    return one_df

syncDfWithPath()

Sync path with existing df. Used to detect new/removed files.

If we currently have just one file (self._filePath) we will trash it and load a folder

Notes

20231230, trying to use this to open a one file window with an exiting h5 file.

Source code in sanpy/analysisDir.py

def syncDfWithPath(self):
    """Sync path with existing df. Used to detect new/removed files.

    If we currently have just one file (self._filePath) we will trash it and load a folder

    Notes
    -----
    20231230, trying to use this to open a one file window with an exiting h5 file.
    """
    self._pool_frame_cache.clear()

    pathFileList = self.getFileList()

    # our currently loaded files
    dfFileList = self._df["File"].tolist()

    # logger.info(f'dfFileList: {dfFileList}')
    # print('    === pathFileList (on drive):')
    # print('    ', pathFileList)
    # print('    === dfFileList (in table):')
    # print('    ', dfFileList)

    addedToDf = False

    # look for files in path not in df
    for pathFile in pathFileList:
        fileName = os.path.split(pathFile)[1]
        if fileName not in dfFileList:
            logger.info(f'   Found file in path "{fileName}" not in df')

            # load bAnalysis and get df column values
            addedToDf = True

            ba, rowDict = self.getFileRow(pathFile)  # loads bAnalysis

            if rowDict is not None:
                # listOfDict.append(rowDict)

                # TODO: get this into getFileROw()
                # logger.warning("bug 20220718, not sure we need this ???")
                # print(rowDict)

                # rowDict['relPath'] = pathFile
                rowDict["_ba"] = None

                self.appendRow(rowDict=rowDict, ba=None)

    # look for files in df not in path
    # for dfFile in dfFileList:
    #     if not dfFile in pathFileList:
    #         logger.info(f'Found file in df "{dfFile}" not in path')

    if addedToDf:
        df = self._df
        df = df.sort_values(
            by=["File"], axis="index", ascending=True, inplace=False
        )
        df = df.reset_index(drop=True)
        self._df = df

    self._updateLoadedAnalyzed()

Functions:

getFileList(path, theseFileTypes, depth=1)

Get a list of files from a path.

Source code in sanpy/analysisDir.py

def getFileList(path, theseFileTypes, depth=1):
    """Get a list of files from a path."""
    fileList = [filePath for filePath in _walk(path, theseFileTypes, depth)]
    return fileList