diff options
| -rw-r--r-- | qolab/tableflow/__init__.py | 32 |
1 files changed, 21 insertions, 11 deletions
diff --git a/qolab/tableflow/__init__.py b/qolab/tableflow/__init__.py index 1466dde..2bd4378 100644 --- a/qolab/tableflow/__init__.py +++ b/qolab/tableflow/__init__.py @@ -16,6 +16,7 @@ import pandas as pd import warnings def loadInOutTables(inputFileName=None, outputFileName=None, comment=None): + """Load input table from a file and if exist output table too, otherwise clone input table to the output one.""" if not inputFileName: return None, None @@ -33,8 +34,11 @@ def loadInOutTables(inputFileName=None, outputFileName=None, comment=None): return tIn, tOut def ilocRowOrAdd(tbl, row): - # Find similar 'row' in 'tbl', NA in both set treated as a hit. - # if similar row not found, insert it. + """Find a row in a table ('tbl') similar to a provided 'row' in 'tbl'. + + NA in both sets treated as a match. + If similar 'row' not found in the table, insert it. + """ tSub = tbl[row.keys()] res = (tSub == row) | (tSub.isna() & row.isna() ) res = res.all(axis=1) # which rows coincide @@ -48,13 +52,17 @@ def ilocRowOrAdd(tbl, row): return i def updateTblRowAt(tbl, i, row): + """Update row with position 'i' in the table ('tbl') with values from 'row'.""" for k in row.keys(): tbl.at[i, k] = row[k] return def isRedoNeeded(row, cols2check): - # redo is required if *all* required entries in cols2check are NA - # or we are missing columns in cols2check list + """Check is Redo required in a given row. + + Redo is required if *all* required entries in 'cols2check' are NA + or we are missing columns in cols2check list + """ for c in cols2check: if c not in row.keys(): return True @@ -63,13 +71,15 @@ def isRedoNeeded(row, cols2check): return False def reflowTable(tIn, tOut, process_row_func=None, postProcessedColums=None, extraInfo=None, redo=False): - # update tOut in place based on the inputs specified in tIn - # effectively maps unprocess rows in to process_row_func - # - postProcessedColums is a list of column names which need to be generated - # - extraInfo is dictionary of additional parameter supplied to process_row_func - # - process_row_func expected to behave like: - # rowOut = process_row_func(rowIn, extraInfo=userInfo) - # - redo controls if reflow is needed unconditionally (i.e. force reflow) + """Reflow/update tOut in place based on the inputs specified in tIn. + + Effectively maps unprocess rows to 'process_row_func' + - postProcessedColums is a list of column names which need to be generated + - extraInfo is dictionary of additional parameter supplied to process_row_func + - process_row_func expected to behave like: + rowOut = process_row_func(rowIn, extraInfo=userInfo) + - redo controls if reflow is needed unconditionally (i.e. force reflow) + """ if not process_row_func: warnings.warn("process_row_func is not provided, exiting reflowTable") return |