python regular expression substitution with matched group

Question

I'm trying to substitue the channel name for AndroidManifest.xml to batch generate a groups of channel apk packages for release.

<meta-data android:value="CHANNEL_NAME_TO_BE_DETERMINED" android:name="UMENG_CHANNEL"/> from an xml file.

The channel configs are saved in a config file, sth like:

channel_name    output_postfix  valid 
"androidmarket" "androidmarket" true

Here is what I tried:

manifest_original_xml_fh = open("../AndroidManifest_original.xml", "r")
manifest_xml_fh = open("../AndroidManifest.xml", "w")
pattern = re.compile('<meta-data\sandroid:value=\"(.*)\"\sandroid:name=\"UMENG_CHANNEL\".*')
for each_config_line in manifest_original_xml_fh:
    each_config_line = re.sub(pattern, channel_name, each_config_line) 
    print each_config_line

It replaces the whole <meta-data android:value="CHANNEL_NAME_TO_BE_DETERMINED" android:name="UMENG_CHANNEL"/> to androidmarket which is obviously not my need. Then I figured out the problem is that pattern.match(each_config_line) return a match result ,and one of the result group is "CHANNEL_NAME_TO_BE_DETERMINED". I've also tried to give some replace implementation function, but still failed.

So, since I've successfully find the pattern, how can I replace the matched result group element correctly?

I don't now the process you want to do, so what is your expected output? I think your misunderstanding is, everything that has been matched will be replaced. If you want to keep stuff from the pattern, you have to capture it and reinsert it in the replacement string. (Or don't match that stuff by using lookaround assertions) — stema, May 04 '12 at 07:21
To make the prob more clear, I'd like to mention this: CHANNEL_NAME_TO_BE_DETERMINED is dynamic here, and what I need is to replace it to some configed channel name. — David, May 04 '12 at 07:26

score 1 · Answer 1 · answered May 04 '12 at 07:28

I suggest a different approach: save your xml as a template, with placeholders to be replaced with standard Python string operations.

E.g.

AndroidManifest_template.xml:

<meta-data android:value="%(channel_name)s" android:name="UMENG_CHANNEL"/>

python:

manifest_original_xml_fh = open("../AndroidManifest_template.xml", "r")
manifest_xml_fh = open("../AndroidManifest.xml", "w")
for each_config_line in manifest_original_xml_fh:
    each_config_line = each_config_line % {'channel_name': channel_name}
    print each_config_line

Good workaround. Actually, this is what I did. I'm just very curious on how to solve it directly. — David, May 04 '12 at 07:48

Joanna Derks · Answer 2 · 2012-05-04T07:22:36.377

To capture just the value of the meta-data tag you need to change the regex:

<meta-data\sandroid:value=\"([^"]*)\"\sandroid:name=\"UMENG_CHANNEL\".*

Specifically I changed this part:

\"(.*)\" - this is a greedy match, so it will go ahead and match as many characters as possible as long as the rest of the expression matches

to

\"([^"]*)\" - which will match anything that's not the double quote. The matching result will still be in the first capturing group

If you want to do the replace thing, a better idea might be to capture what you want to stay the same - I'm not a python expert but something like this would probably work:

re.sub(r'(<meta-data\sandroid:value=\")[^"]*(\"\sandroid:name=\"UMENG_CHANNEL\".*)'
, r'\1YourNewValue\2', s)

\1 is backreference 1 - i.e. it gets what the first capturing group matched

:Thanks for the reply. The match I did seems to be no prob. The matched result group element is CHANNEL_NAME_TO_BE_DETERMINED, all I wanna do is replace this matched result group element to some channel name. — David, May 04 '12 at 07:25

score 0 · Accepted Answer · answered May 04 '12 at 07:32

I think your misunderstanding is, everything that has been matched will be replaced. If you want to keep stuff from the pattern, you have to capture it and reinsert it in the replacement string.

Or match only what you want to replace by using lookaround assertions

Try this

pattern = re.compile('(?<=<meta-data\sandroid:value=\")[^"]+')
for each_config_line in manifest_original_xml_fh:
    each_config_line = re.sub(pattern, channel_name, each_config_line)

(?<=<meta-data\sandroid:value=\") is a positive lookbehind assertion, it ensures that this text is before, but does not match it (so it will not be replaced)

[^"]+ will then match anything that is not a "

See it here on Regexr

Very concise and effective solution. Never heard of positive lookbehind assertion before, have learned it now. Thanks for that. — David, May 04 '12 at 07:59

python regular expression substitution with matched group

3 Answers3