U
|
¬ý°d' ã @ s¦ d dl mZ d dlmZmZ d dlZd dlmZm Z m
|
Z
|
m Z d dl m Z d dlmZ d dlZd dlmZ d dlmZ erd d lmZ eed
|
d G d d d eZdS )é )Ú annotations)Ú TYPE_CHECKINGÚcastN)ÚFilePathÚ
|
ReadBufferÚScalarÚStorageOptions)Úimport_optional_dependency)Údoc)Ú _shared_docs)ÚBaseExcelReader)ÚNaTTypeÚstorage_options©r c sÚ e Zd Zd,dddd fdd Zedd Zdd
|
d d Zed dddZeddddZddddZ d dddZ
|
d-dd d d d Z ddd d!Z ddd"d#Z d$dd%d&Zd'dd(d)Zd dd*d+Z ZS ).Ú ODFReaderNz FilePath | ReadBuffer[bytes]r ÚNone)Úfilepath_or_bufferr Úreturnc s t d t j||d dS )zß
|
Read tables out of OpenDocument formatted files.
|
|
Parameters
|
----------
|
filepath_or_buffer : str, path to be parsed or
|
an open readable stream.
|
{storage_options}
|
Zodfr N)r ÚsuperÚ__init__)Úselfr r ©Ú __class__© úQd:\z\workplace\vscode\pyvenv\venv\Lib\site-packages\pandas/io/excel/_odfreader.pyr s zODFReader.__init__c C s ddl m} |S )Nr )Ú OpenDocument)Úodf.opendocumentr )r r r r r Ú_workbook_class/ s zODFReader._workbook_class)r c C s ddl m} ||S )Nr )Úload)r r )r r r r r r Ú load_workbook5 s zODFReader.load_workbookÚstr)r c C s dS )z'Property for compat with other readers.Ú r )r r r r Ú empty_value: s zODFReader.empty_valuez list[str]c C s& ddl m} | j |¡}dd |D S )z4Return a list of sheet names present in the documentr ©ÚTablec S s g | ]}| d ¡qS ©Úname)Ú getAttribute)Ú.0Útr r r Ú
|
<listcomp>E s z)ODFReader.sheet_names.<locals>.<listcomp>)Ú odf.tabler$ ÚbookÚgetElementsByType)r r$ Útablesr r r Ú sheet_names? s zODFReader.sheet_namesÚint)Úindexc C s* ddl m} | |¡ | j |¡}|| S )Nr r# )r+ r$ Zraise_if_bad_sheet_by_indexr, r- )r r1 r$ r. r r r Úget_sheet_by_indexG s
|
z ODFReader.get_sheet_by_indexr% c C s^ ddl m} | |¡ | j |¡}|D ]}| d¡|kr&| S q&| ¡ td| dd S )Nr r# r& zsheet z
|
not found)r+ r$ Zraise_if_bad_sheet_by_namer, r- r' ÚcloseÚ
|
ValueError)r r& r$ r. Útabler r r Úget_sheet_by_nameN s
|
|
zODFReader.get_sheet_by_namez
|
int | Nonez list[list[Scalar | NaTType]])Úfile_rows_neededr c s ddl m}m}m} | j}| j}||h | |¡}d} d}
|
g } |D ]
|
} fdd| jD } d}g }| D ]d}|j|kr| |¡}n| j}| |¡}|| jkrª||7 }ql|
|
| jg| ¡ d}|
|
|g| ¡ ql|
|
t |k ræt |}
|
| | ¡}| | ¡r| |7 } n2|
|
| jgg| ¡ d} t|D ]}| |¡ q&|dk rFt | |krF qTqF| D ].}t ||
|
k rX|
|
| jg|
|
t | ¡ qX| S )z9
|
Parse an ODF Table into a list of lists
|
r )ÚCoveredTableCellÚ TableCellÚTableRowc s$ g | ] }t |d r|j kr|qS )Úqname)Úhasattrr; )r( Úx©Ú
|
cell_namesr r r* r s
|
|
þz,ODFReader.get_sheet_data.<locals>.<listcomp>N)r+ r8 r9 r: r; r- Ú
|
childNodesÚ_get_cell_valuer" Ú_get_column_repeatÚextendÚlenÚ_get_row_repeatÚ _is_empty_rowÚrangeÚappend)r Zsheetr7 r8 r9 r: Zcovered_cell_nameZtable_cell_nameZ
|
sheet_rowsZ
|
empty_rowsZ max_row_lenr5 Z sheet_rowZ sheet_cellsZ empty_cellsZ table_rowZ
|
sheet_cellÚvalueZ column_repeatZ
|
row_repeatÚ_Úrowr r> r Úget_sheet_data[ sN
|
|
|
þ
|
|
|
|
|
|
zODFReader.get_sheet_datac C s" ddl m} t|j |dfd¡S )z«
|
Return number of times this row was repeated
|
Repeating an empty row appeared to be a common way
|
of representing sparse rows in the table.
|
r ©ÚTABLENSznumber-rows-repeatedé ©Úodf.namespacesrN r0 Ú
|
attributesÚget)r rK rN r r r rE s zODFReader._get_row_repeatc C s" ddl m} t|j |dfd¡S )Nr rM znumber-columns-repeatedrO rP )r ÚcellrN r r r rB ª s z ODFReader._get_column_repeatÚboolc C s$ |j D ]}t|j dkr dS qdS )z4
|
Helper function to find empty rows
|
r FT)r@ rD )r rK Úcolumnr r r rF ¯ s
|
zODFReader._is_empty_rowzScalar | NaTTypec C sB ddl m} t|dkr tjS |j |df¡}|dkrJt|dkrFdS dS |d krX| jS |d krt|j |d
|
f¡}t |}||kr|S |S |d kr¬|j |d
|
f¡}t|S |d kr¾|
|
|¡S |d krÞ|j |d
|
f¡}t|S |dkr|j |df¡}t |¡S |dkr(t t|¡}t t| ¡ S | ¡ td| d S )Nr )ÚOFFICENSz#N/Az
|
value-typeÚbooleanÚTRUETFÚfloatrI Ú
|
percentageÚstringÚcurrencyÚdatez
|
date-valueÚtimezUnrecognized type )rQ rW r ÚnpÚnanrR rS r" rZ r0 Ú_get_cell_string_valueÚpdÚ Timestampr r r_ r3 r4 )r rT rW Z cell_typeZ
|
cell_valueÚvalZstampr r r rA ¹ s@
|
|
|
|
zODFReader._get_cell_valuec C s¦ ddl m} ddlm} ddlm} | j}g }|jD ]d}t||r|j|krtt |j
|
|dfd¡}| d| ¡ q| | |¡¡ q6| t| d¡¡ q6d |¡S )
|
z
|
Find and decode OpenDocument text:s tags that represent
|
a run length encoded sequence of space characters.
|
r )ÚElement)ÚTEXTNS)ÚSÚcrO ú Ú
|
r! )Z odf.elementrf rQ rg Zodf.textrh r; r@ Ú
|
isinstancer0 rR rS rH rb r ÚstripÚjoin) r rT rf rg rh Ztext_srI ÚfragmentÚspacesr r r rb à s
|
|
|
z ODFReader._get_cell_string_value)N)N)Ú__name__Ú
|
__module__Ú __qualname__r Úpropertyr r r" r/ r2 r6 rL rE rB rF rA rb Ú __classcell__r r r r r s$ ý
|
ÿE
|
|
'r )Ú
|
__future__r Útypingr r Únumpyr` Zpandas._typingr r r r Zpandas.compat._optionalr Zpandas.util._decoratorsr
|
Zpandasrc Zpandas.core.shared_docsr Zpandas.io.excel._baser Zpandas._libs.tslibs.nattyper r r r r r Ú<module> s
|
