HR-eF b d Z ddlZddlmZ ddlmZ ddlmZ ddlm Z G d d e
Z G d
d Z d Z
G d
de j Z G d de j Z G d de j Z G d de j Z G d de j Z G d de j ZdS )zAn extensible HTML table reader and writer.
html.py:
Classes to read and write HTML tables
`BeautifulSoup `_
must be installed to read HTML tables.
N)deepcopy)Column)writer )corec e Zd ZdZd Zd ZdS )
SoupStringz8
Allows for strings to hold BeautifulSoup data.
c , t j | g|R i |S N)str__new__)clsargskwargss 5lib/python3.11/site-packages/astropy/io/ascii/html.pyr
zSoupString.__new__ s" {30000000 c || _ d S r )soup)selfvals r __init__zSoupString.__init__ s
r N)__name__
__module__ __qualname____doc__r
r r r r r s< 1 1 1 r r c e Zd ZdZd Zd ZdS )
ListWriterzD
Allows for XMLWriter to write to a list instead of a file.
c || _ d S r )out)r r s r r zListWriter.__init__% s
r c : | j | d S r )r append)r datas r writezListWriter.write( s r N)r r r r r r$ r r r r r s< r r c | | j dk rdS d|vr|dk S |d }t |t rd| j v o | d |k S t |t r||k S dS )zb
Checks whether the given BeautifulSoup tag is the table
the user intends to process.
NtableFtable_idr id)name
isinstancer attrsint)r htmldictnumtabler' s r identify_tabler/ , s
|tyG++u 8 # #1}
#H(C $tz!`_ to be installed.
c ddl m} n"# t $ r t j d w xY wd| j vr_t
j 5 t
j dd |d | }ddd n # 1 swxY w Y n* |d | | j d }|
d }t | D ]"\ }}t || j |d
z r|} n[#t | j d t rd | j d }nd
| j d d}t j d| d d |
d D } | S )zh
Convert the given input into a list of SoupString rows
for further processing.
r )
BeautifulSoupz3BeautifulSoup must be installed to read HTML tablesparserignorez&.*no parser was explicitly specified.*
Nr& r r' znumber zid ''zERROR: HTML table z
not foundc , g | ]}t | S r )r .0xs r
z.HTMLInputter.process_lines..m s AAAqZ]]AAAr tr)bs4r3 ImportErrorr OptionalTableImportErrorhtmlwarningscatch_warningsfilterwarningsjoinfind_all enumerater/ r* r, InconsistentTableError)
r linesr3 r tablesipossible_tabler& err_descr soup_lists
r
process_lineszHTMLInputter.process_linesI s
) ) ) ) ) ) ) /E
49 $ $(**
7
7' F %}TYYu%5%566
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7 !=5!1!149X3FGGD w''!*6!2!2 A~ ndiQ??
&
$)J/55
< =di
&; = = ;49Z#8 ; ; ; -:Y:::
BAENN4,@,@AAA s (4BB B N)r r r r rO r r r r1 r1 A s- & & & & &r r1 c e Zd ZdZd ZdS ) HTMLSplitterz
Split HTML table data.
c # L K |D ]w}t |t st d |j }| d }|rd |D V | d }|rd |D V xt | dk rt
j d dS ) z=
Return HTML data from lines as a generator.
'HTML lines should be of type SoupStringthc g | ]Q}| d r!|j |d fn|j RS colspan)has_attrtextstripr: els r r< z)HTMLSplitter.__call__.. s` {{9--)RW]]__bm44 r tdc @ g | ]}|j S r )rY rZ r[ s r r< z)HTMLSplitter.__call__.. s" ???2rw}}???r r z.HTML tables must contain data in a tagN)r* r TypeErrorr rF lenr rH )r rI liner header_elements
data_elementss r __call__zHTMLSplitter.__call__w s @ @D dJ//
K IJJJ9D "mmD 1 1O
. !MM$ / /M
@?? ?????? u::??-@
?r N)r r r r rd r r r rQ rQ r s- r rQ c e Zd ZdZ ej e ej e ej e gZ fdZ
xZ S )
HTMLOutputterz
Output the HTML data as an ``astropy.table.Table`` object.
This subclass allows for the final table to contain
multidimensional columns (defined using the colspan attribute
of | ).
c g }d}|t | k r|| }t |d rp||||j z }t j |j }t
t d |D |_ | | ||j z
}n| | |dz
}|t | k t || S )z?
Process the data in multidimensional columns.
r rW c g | ] }|j
S r )str_valsr9 s r r< z*HTMLOutputter.__call__.. s -L-L-LQaj-L-L-Lr r ) r` hasattrrW r r r) listzipri r" superrd ) r colsmetanew_colscol_numcol span_colsnew_col __class__s r rd zHTMLOutputter.__call__ s D !!w-CsI&&
7S[+@!@ A +ch//#'-L-L)-L-L-L(M#N#N (((3;&$$$1 D !! ww $///r ) r r r r r
convert_numpyr, floatr default_convertersrd
__classcell__ru s @r rf rf s| 3 5!!3 0 0 0 0 0 0 0 0 0r rf c e Zd ZeZd Zd ZdS )
HTMLHeaderc t | D ];\ }}t |t st d |j }|j |c S
tag and will define stylistic attributes of the output.
* table_id : ID for the input table
If a string, this defines the HTML id of the table to be processed.
If an integer, this specifies the index of the input table in the
available tables. Unless this parameter is given, the reader will
use the first table found in the input file.
* multicol : Use multi-dimensional columns for output
The writer will output tuples as elements of multi-dimensional
columns if this parameter is true, and if not then it will
use the syntax 1.36583e-13 .. 1.36583e-13 for output. If not
present, this parameter will be true by default.
* raw_html_cols : column name or list of names with raw HTML content
This allows one to include raw HTML content in the column output,
for instance to include link references in a table. This option
requires that the bleach package be installed. Only whitelisted
tags are allowed through for security reasons (see the
raw_html_clean_kwargs arg).
* raw_html_clean_kwargs : dict of keyword args controlling HTML cleaning
Raw HTML will be cleaned to prevent unsafe HTML from ending up in
the table output. This is done by calling ``bleach.clean(data,
**raw_html_clean_kwargs)``. For details on the available options
(e.g. tag whitelist) see:
https://bleach.readthedocs.io/en/latest/clean.html
* parser : Specific HTML parsing library to use
If specified, this specifies which HTML parsing library
BeautifulSoup should use as a backend. The options to choose
from are 'html.parser' (the standard library parser), 'lxml'
(the recommended parser), 'xml' (lxml's XML parser), and
'html5lib'. html5lib is a highly lenient parser and therefore
might work correctly for unusual input if a different parser
fails.
* jsfiles : list of js files to include when writing table.
* cssfiles : list of css files to include when writing table.
* js : js script to include in the body when writing table.
* table_class : css class for the table
rA z.htmlz
HTML table c t t | | _ d|vr
d| j d< d|vr
d| j d< | j | j _ dS )zB
Initialize classes for HTML reading and writing.
multicolTr' r N)rm r r rA inputter)r r- ru s r r z
HTML.__init__J sj X&& X % %$(DIj ! X % %$%DIj ! !Y
r c l t | _ t | S )zU
Read the ``table`` in HTML format and return a resulting ``Table``.
)rf outputterrm read)r r& ru s r r z HTML.readV s& ' ww||E"""r c | | t |j }|| j j _ || j _ t | j j t r| j j g| j _ | j
| | j g }| j
dg t t rgfd|D }| j
di }t j t# | }| d 5 | d 5 | dddi 5 d
d
d
n # 1 swxY w Y | dd d d
5 d
d
d
n # 1 swxY w Y d| j v rM| d 5 | | j d d
d
d
n # 1 swxY w Y d| j v rB| j d D ]4}| dd|d 5 d
d
d
n # 1 swxY w Y 5d| j v rT| j d D ]F}| d| 5 | d d
d
d
n # 1 swxY w Y Gd
d
d
n # 1 swxY w Y | d 5 d| j v rz| d 5 | d 5 | | j d d
d
d
n # 1 swxY w Y d
d
d
n # 1 swxY w Y t | j d t r| j d }nd
}d | j v r| j d } d | i}
ni }
| d ||
5 | d! 5 | d" 5 |D ]} t) | j d#k r0| j d$ r#| d%| j d# & n| d% | | j j | d'( d
d
d
n # 1 swxY w Y g } g }
g }t7 || D ])\ } }t) | j d#k r| j d$ r| j d# }t9 | D ]t; fd)| D }| | |j }| | |
| | | | | | j }| | |
| + d
d
d
n # 1 swxY w Y t7 | D ]}| d" 5 t7 ||
D ]\ }}|rd*nd+} |j |fi |5 | d, | | | d'( d
d
d
n # 1 swxY w Y d
d
d
n # 1 swxY w Y d
d
d
n # 1 swxY w Y d
d
d
n # 1 swxY w Y d
d
d
n # 1 swxY w Y d | gS )-zR
Return data in ``table`` converted to HTML as a list of strings.
raw_html_colsc , g | ]}|j j vS r )infor) )r: rr r s r r< z HTML.write..u s" KKKs
]:KKKr raw_html_clean_kwargsrA headro charsetzutf-8)attribNz Content-typeztext/html;charset=UTF-8)z
http-equivcontentcssstylecssfileslink
stylesheetztext/css)relhreftypejsfilesscript)srcr bodyjsnoner' table_classclassr& )r( r theadr= r r rT rV F)indentc g | ]
}| S r r )r: r\ rK s r r< z HTML.write.. s 5J5J5Jbe5J5J5Jr
escape_xml bleach_cleanr] )!_check_multidim_tablerk columnsvaluesr# headerrn r* fill_valuesr _set_fill_values_set_col_formatsrA getr r XMLWriterr tagxml_cleaning_methodr` shapestartr r) rZ endrl r r
iter_str_valsr" rE )r r& rn rI cols_escapedr wfilename
html_table_idhtml_table_classr rr
col_str_itersnew_cols_escapedrp col_escapedspanrt new_col_iter_str_valscol_iter_str_valsrowr\ methodrK r s @@r r$ z
HTML.write] sM
""5)))EM((**++ $ di+U 3 3 <%)Y%:$;DI ! ""4((( ""$$$
or::
mS ) ) , *OMKKKKdKKK !%
.Er J J
Z . . / /
UU6]] ] 8 ] 8v
'
'UU69g*>U?? UU&4#< DI%%w 1 1ty /0001 1 1 1 1 1 1 1 1 1 1 1 1 1 1 **$(Ij$9 ! ! UU " 8* # ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ))$(Ii$8 ' ' UU8U : : ' ' FF2JJJ' ' ' ' ' ' ' ' ' ' ' ' ' ' '/
'
'
'
'
'
'
'
'
'
'
'
'
'
'
'4 v B
8 B
849$$..v66 4 4 UU8__ 4 4 FF49T? 3 3 34 4 4 4 4 4 4 4 4 4 4 4 4 4 44 4 4 4 4 4 4 4 4 4 4 4 4 4 4 di
3S99 )$(Ij$9MM$(M DI--'+y '?$ %'7 8FF FUU7}VUDD 48 48w 'E 'E UU4[[ 4 4'+ 4 4#&sy>>A#5#5$)J:O#5$%GGD#)A,G$G$G$G$G$%GGDMMM !sx}':':'<'< = = = !U 3 3 3 3 44 4 4 4 4 4 4 4 4 4 4 4 4 4 4 )+
+-( $&03D,0G0G E E ,C "39~~ 1 1di
6K 1'*y|).t != !=A.45J5J5J5Jc5J5J5J.K.KG<@ \ M73!M M7 M$
$M7'M$
(M7+ \ 7M; ;\ >M; ?A(\ '\ =X# B*S
> X#
SX# SEX# \ #X'
'\ *X'
+#\ *[0 8A[ [0 [[0 [ [0 $ \ 0[44\ 7[48\ = \ \
\ \
\ \7 \$ $\7'\$ (\77\;>\;c # K t |d }t |d }t | D ]O\ }}|r)|r'|j | r|j t j V 0|r||j v r|j | V K|V PdS )zZ
Return an iterator of the values with replacements based on fill_values.
maskr N)rj rG r r r masked)r rr r is_masked_columnhas_fill_valuesidxcol_strs r r zHTML.fill_values s
#3//!#}55 %m 4 4 LC
O
8C= /$+6666
co-- /'2222 MMMM r )r r r r _format_name_io_registry_format_aliases_io_registry_suffix _descriptionr| header_classr
data_classr1 inputter_classmax_ndimr r r$ r ry rz s @r r r s 6 6p L#)( ! L LJ!NH "
'
'
'
'
'
'# # # # #@ @ @ D r r )r rB copyr
astropy.tabler astropy.utils.xmlr r r r r r r/ BaseInputterr1 BaseSplitterrQ TableOutputterrf
BaseHeaderr| BaseDatar
BaseReaderr r r r r s $ $ $ $ $ $ *. . . . .4$ . . .b 4$ >#0 #0 #0 #0 #0D' #0 #0 #0L% % % % % % % % P$ $ $ $ $ t} $ $ $ Nl l l l l4? l l l l lr |