使用Python提取HTML标签内容

Question

I have a Word document running to 188 pages which uses mainly font sizes to denote structure.

You can see it here: https://github.com/watty62/jazz_birthdays/blob/master/jazz_birthdays.doc

Using Python (my preferred language) I would like to extract the content and save it to a data format such as json.

I opened the doc in Libre Office and saved it as HTML and also tried exporting it as an alternative XML file.

You can see the XMl and HTML files here Both seem to produce reasonably structured docs but extracting the meaning from the XML is more difficult

<para>1 January   </para>
<para>Helmut Brandt, baritone sax, 1931 (July 26, 2001)</para> 

In the HTML version we end up with

    <P LANG="en-US" STYLE="margin-top: 0.18cm; margin-bottom: 0.18cm; page-break-after: avoid">
<FONT SIZE=4>1 January   </FONT>
</P>
<P LANG="en-US" CLASS="western" STYLE="font-weight: normal">Helmut
Brandt, baritone sax, 1931 (July 26, 2001)</P>

Each date is encased in <FONT SIZE=4> </FONT> tags (although these are used occasionally for other purposes.

A quick count give 377 uses of <FONT SIZE=4> - so assuming for now that all 366 days of the year are there then there are 11 uses of it which I'll have to ignore.

My approach was to be to replace the first <Font size=4> with something to denote that it is the opening of the date field, eg <Date> then each subsequent one with a closing of the date (after all the musicians with that birthday) and open the next date thus </Date><Date>

After that I thought that I'd simplify each line - although these will get complicated with name (possibly containing a nickname), instruments played separated by commas, year of birth, and date of death (in brackets and starting "d.") - so lots more to get stuck into later.

An initial attempt with Beautiful soup to parse the file threw up some encoding errors in the original file.

I'm not looking for a solution (as it is a real biggie) but would appreciate any prompts on approach, libraries etc to get me started please.

Thanks

Ian

Answer 1

I hope that this is what you are looking for (if it isn't then please let me know so that I can remove my answer for you):

import re
s="""<P LANG="en-US" STYLE="margin-top: 0.18cm; margin-bottom: 0.18cm; page-break-after: avoid">
<FONT SIZE=4>1 January   </FONT>
</P>
<P LANG="en-US" CLASS="western" STYLE="font-weight: normal">Helmut
Brandt, baritone sax, 1931 (July 26, 2001)</P>"""
print re.findall(r"\d{1,2} \w+",s)

This Outputs:

['1 January']

As a quick explanation the re module is a fancy searching mechanism. It's final() method takes a pattern to search for and a string to search in. I fed it the pattern r"\\d{1,2} \\w+". The r before the string tells python to ignore the backslashes so that re can use them for it's purposes. \\d means a digit. {1,2} means one or two times. The space just means a space. \\w means a word-character. And + means one or more of.