index.htmlを解析してメソッド一覧を作る。フレームに依存しているため、フレームがないjavadocだと落ちそう。

gen_methodlist.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
 
import os
import os.path
import sys
 
import lxml.html
import xlwt
 
def is_method(method):
    if method.find('-') != -1:
        return True
    if method.find('(') == -1:
        return False
    if method.find(')') == -1:
        return False
    return True
 
def write_on_sheet(doc_file, sheet, start, xpath_exp='//a/@name', debug=True):
    root = lxml.html.parse(doc_file)
    methods = [method for method in root.xpath(xpath_exp) if is_method(method)]
    methods.sort()
 
    # write
    sheet.write(start, 1, os.path.basename(doc_file.split('.')[-2])) # remove extention
    for method in methods:
        method_sig = method
        if method.find('-'):
            if (method.startswith('Z:Z')):
                method = method[len('Z:Z'):]
            method_sig = method.split('-')[0] + '(' + ','.join([param.replace(':A', '[]') for param in method.split('-')[1:-1]]) + ')'
        if debug:
            print method_sig
        sheet.write(start, 2, method_sig)
        start += 1
    return start
    
def main():
    if len(sys.argv) != 2:
        print 'Error'
        sys.exit(1)

    package_doc_files = {}

    root_addr = sys.argv[1][:-len("index.html")] if sys.argv[1].endswith("index.html") else sys.argv[1]
    overview = lxml.html.parse(sys.argv[1]) # sys.argv[1] = index.html
    package_list_frame_addr = root_addr + overview.xpath('//*[@name="packageListFrame"]')[0].xpath('@src')[0]
    package_list_frame = lxml.html.parse(package_list_frame_addr)
    package_overview_list = [package_overview for package_overview in package_list_frame.xpath('//a[@target="packageFrame" and contains(@href, "/package-frame.html")]')]
    for package_overview_link in package_overview_list:
        package_overview_addr = root_addr + package_overview_link.xpath('@href')[0]
        package_overview = lxml.html.parse(package_overview_addr)
        class_list = [class_overview.xpath("@href")[0] for class_overview in package_overview.xpath('//a[@target="classFrame" and not(contains(@href, "/package-summary.html"))]')]
        class_addr_list = [(package_overview_addr[:-len("package-frame.html")] + class_addr) for class_addr in class_list]
        package_doc_files.update({package_overview_link.text:class_addr_list})

    # excel file
    book_name = 'method_list.xls'
    sheet_name = 'All Methods'
    if os.path.exists(book_name):
        try:
            os.remove(book_name)
        except ex:
            print "Sheet can't be removed"
            print ex
            sys.exit(1)
 
    wb = xlwt.Workbook()
    ws = wb.add_sheet(sheet_name, cell_overwrite_ok=True)
 
    start = 0
    package_name_list = package_doc_files.keys()
    package_name_list.sort()
    for package_name in package_name_list:
        print package_name
        ws.write(start, 0, package_name)
        for doc_file in package_doc_files[package_name]:
            print doc_file
            start = write_on_sheet(doc_file, ws, start)
 
    wb.save(book_name)
 
if __name__ == '__main__':
    main()