Extracting data between two tags in HTML file

问题:

I've got a HUUUGE HTML file here saved on my system, which contains data from a product catalogue. The data is structured such that for each product record the name is between two tags (name) and (/name) .

Each product has up to 3 attributes: name, productID, and color, but not all products will have all these attributes.

How would I go about extracting this data for each product without mixing up the product attributes? The file is also 50 megabyte!

Code example ....

<name>'hat'</name>
blah blah blah
<prodId>'1829493'</prodId>
blah blah blah
<color>'cyan'</color>

blah blah 
blah blah blah
blah blah blah

<name>'shirt'</name>
blah blah blahblah blah blah
<prodId>'193'</prodId>

<name>'dress'</name>
blah blah blah
blah blah blah
<prodId>'18'</prodId>
<color>'dark purple'</color>

回答1:

A file of size 50 MB isn't so big that you can't just load its contents directly into MATLAB as a string, which you can do with the function FILEREAD:

strContents = fileread('yourfile.html');

Assuming the file format you have above, you can then parse the contents with the function REGEXP (using named token capture):

expr = '<(?<tag>name|prodId|color)>''([^<>]+)''</\k<tag>>';
tokens = regexp(strContents,expr,'tokens');
tokens = vertcat(tokens{:});

And the contents of token using your sample file contents will be:

tokens = 

    'name'      'hat'        
    'prodId'    '1829493'    
    'color'     'cyan'       
    'name'      'shirt'      
    'prodId'    '193'        
    'name'      'dress'      
    'prodId'    '18'         
    'color'     'dark purple'

You may then want to parse the resulting N-by-2 cell array and place the contents in a structure array with fields 'name', 'prodId', and 'color'. The difficulty is that not every entry will have all three fields. Assuming each 'name' will be followed by either a 'prodId', a 'color', or both (in the order 'prodId' then 'color'), then the following code should work for you:

s = struct('name',[],'prodId',[],'color',[]);  %# Initialize structure
nTokens = size(tokens,1);                      %# Get number of tokens
nameIndex = find(strcmp(tokens(:,1),'name'));  %# Find indices of 'name'
[s(1:numel(nameIndex)).name] = deal(tokens{nameIndex,2});  %# Fill 'name' field

%# Find and fill 'prodId' that follows a 'name':
index = strcmp(tokens(min(nameIndex+1,nTokens),1),'prodId');
[s(index).prodId] = deal(tokens{nameIndex(index)+1,2});

%# Find and fill 'color' that follows a 'name':
index = strcmp(tokens(min(nameIndex+1,nTokens),1),'color');
[s(index).color] = deal(tokens{nameIndex(index)+1,2});

%# Find and fill 'color' that follows a 'prodId':
index = strcmp(tokens(min(nameIndex+2,nTokens),1),'color');
[s(index).color] = deal(tokens{min(nameIndex(index)+2,nTokens),2});

And the contents of s using your sample file contents will be:

>> s(1)

      name: 'hat'
    prodId: '1829493'
     color: 'cyan'

>> s(2)

      name: 'shirt'
    prodId: '193'
     color: []

>> s(3)

      name: 'dress'
    prodId: '18'
     color: 'dark purple'

回答2:

There are two ways of solving this sort of problem: string manipulation with regexes (as suggested by gnovice) or parsing the file (or a mix of the two). Parsing is often best if your file is very well structured; regexes win for messy files.

Here's the parsing solution.

Start by downloading xmliotools, and calling xml_read on your file. Your example isn't completely reproducible, so here are two different versions of the data.

Save this to test1.xml:

<?xml version="1.0" encoding="utf-8"?>
<root>
<name>'hat'</name>
<prodId>'1829493'</prodId>
<color>'cyan'</color>
<name>'dress'</name>
<prodId>'18'</prodId>
<color>'dark purple'</color>
</root>

Save this to test2.xml.

<?xml version="1.0" encoding="utf-8"?>
<root>
<item>
<name>'hat'</name>
<prodId>'1829493'</prodId>
<color>'cyan'</color>
</item>
<item>
<name>'dress'</name>
<prodId>'18'</prodId>
<color>'dark purple'</color>
</item>
</root>

Now compare

x1 = xml_read('test1.xml')
x2 = xml_read('test2.xml')