(轉)幾個有趣的python開源項目

一直都對搜尋引擎很感興趣，是以有一個是python的蜘蛛程式。另一個是用python操作pdf格式檔案。

1、用python來做一個蜘蛛程式抓取網頁，有了urllib庫，真是太簡單了。另外網頁的解析也有相應的庫sgmllib可以使用。不過還不知道python的sgmllib有沒有類似 Jtidy 的規範html代碼的功能，或者是有另外的庫來幹這事。

比較有名氣的：

Harvest Man------http://code.google.com/p/harvestman-crawler/

HarvestMan is a modular, extensible and flexible web crawler program cum framework written in pure Python. HarvestMan can be used to download files from websites according to a number of customized rules and constraints. It can be used to find information from websites matching keywords or regular expressions.

The final goal of the project is to develop a full-fledged semantic personal data mining platform which can be used to retrieve information from the Internet in a highly customizable manner, so that one can fetch information from the web the way he wants it, when he wants it. For this, HarvestMan project will provide support for Web 2.0 and 3.0 technologies such as RSS, RDF, OWL etc. （這個目标還真是大啊，要是真的可以做到那就真是牛逼。）

另外，還有一些小的項目，用Google code或者 sourceforge.net搜尋，就可以找到。

舉個例子：

http://code.google.com/p/supercrawler/

2、對pdf檔案的操作，C++,c#和java都有一些開源的類庫可以使用。比如：pdflib,itext,pdfclown,pdfbox.

他們可以實作pdf檔案的解析，并實作pdf與rtf html xml等格式之間的互相轉換。

今天發現了一個可以操作pdf的python庫： pdfminer.

http://code.google.com/p/pdfminerr/

不知道有沒有其他的庫。希望高手補充。

3、有了pdf的操作庫，可以輕松實作對pdf檔案内容的有目的的抽取。

這裡有一個抽取文獻的參考文獻的例子：

pdf2ref

http://code.google.com/p/pdftoref/

This project aims to develop an efficient rule based extractor of entries of references, located in scientific articles in English language. The application takes a pdf file or a directory of pdf and then returns an html file, containing the list of all entries with their respective title. Moreover the title of the article cited is searched through Google Web Service to get the URL that identifying the article on the web. If the URL provides on the page a Bibtex entry, this will appear in the html output under the relative entries, stolen from some typical site like citeseer, ieeexlpore etc. The application does not make search over pdf file based on images.

Result

ImplementationPdfminer tool, a pure python lib to manage pdf; instead to query Google Web Service it uses Python SoapPy. This project is born into the course of "Data Base II and Information Retreival" of Master Computer Science at the University of Florence from two engineers. The project is released under the GNU General Public License.

In this way, the best result will be the html files with all entries with their rispective title linked to the URI and below the Bibtex entry. See example below.

The tool is implemented through python scripting and created to run on GNU/Linux. It uses as backend

UsageExamples of UsageExamples of OutputThe article in pdf text format

[email protected]:~$ pdftoref --help
     Usage: 

     -h            --help              This help Command 
     -f filepath   --file=filepath     Run on a file 
     -d directory  --dir=directory     Run into a directory 
     -u            --url               Get the title url 
     -b            --bibtex            Get the bibtex article 
 
     Examples:

     pdftoref -b -u -f ~/file.pdf
     pdftoref --url --bibtex --file=~/file.pdf
     pdftoref -d /home/user/articles 
      

Extract the referencies from a file

us[email protected]:~$pdftoref -u -b -f /home/lizardking/path/to/file.pdf      

The output will be:

PdftoRef> Running on the file: /home/lizardking/path/to/file.pdf
* Extracting the text from pdf      [Done]
* Extracting the entries and titles      [Done]
* Querying Google and writing down html file      [Done]
PdftoRef> Finished file: /home/lizardking/path/to/file.pdf
      

Then run to see result:

[email protected]:~$firefox /home/lizardking/path/to/file.html      

Almost the same for extract the referencies from files in a directory, except that it will use on all pdf files.

[email protected]:~$pdftoref -d /home/lizardking/path/to/dir/      

• Here is the result of the parsing