"""Full-text indexing module for aether

In your _code.py, insert something like
    from search import *
and modify the location of the swish binary SWISH = "..."

You'll also need to set up the indexing process, including the use of
"aetherfilter", a C program that parses aether-markup files.
"""


SWISH = "/usr/local/bin/swish-e"

import os, sys, time, string
from __main__ import *

__all__ = ['handle_search', 'handle_sitemap']

try; enumerate
except NameError:
    def enumerate(i): return zip(range(len(i)), i)

def age(f):
    seconds = time.time() - os.stat(f).st_mtime
    if seconds < 60: return "less than a minute"
    if seconds < 120: return "1 minute"
    if seconds < 60 * 60: return "%d minutes" % ((seconds + 30)/60)
    if seconds < 7200: return "1 hour"
    if seconds < 24 * 60 * 60: return "%d hours" % ((seconds + 30*60)/60/60)
    if seconds < 2*86400: return "1 day"
    return "%d days" % ((seconds + 30*60*24)/60/60/24)

from sgmllib import SGMLParser

class HTMLTextExtractor(SGMLParser):
    def __init__(self):
        SGMLParser.__init__(self)
        self.data = []

    def handle_data(self, data):
        if data: self.data.append(data)

def html2text(document):
    h = HTMLTextExtractor()
    h.feed(document)
    return " ".join(h.data)


def handle_sitemap(name, query):
    search_text = query.get('search',u'')
    search = search_text.lower().split()

    result = [ ]

    def helper(name):
        if exists(name):
            text = load(name).lower()

            for term in search:
                if text.find(term) == -1:
                    break
            else:
                result.append(u'[line [link ' + (quote_markup(name) or u'/') + 
                              u'[page '+quote_markup(name)+']]]')

        names = list_names(name)
        for name, full_name in names:
            helper(full_name)
            

    helper('')

    if not result:
        result.append('Search found nothing, sorry.')

    if search:
        result.insert(0, u'[title Search results]\n\n')
        result.insert(1, u'Searching for: [bold ' + quote_markup(string.join(search,' ')) + u']\n\n')
    else:
        result.insert(0, u'[title Site map]\n\n')

    return make_http_page(string.join(result,u''))


def popen(*args):
    reader, writer = os.pipe()
    p = os.fork()
    if p == 0:
        os.close(reader)
        os.dup2(writer, 1)
        os.close(writer)
        os.execvp(args[0], args)
        os._exit(99)
    os.close(writer)
    return os.fdopen(reader)

def swish(db, term):
    p = popen(SWISH, "-x", "%p\n", "-f", db, "-w", term)
    result = []
    for line in p:
        line = line.strip()
        if line.startswith("#"): continue
        if line == ".": continue
        result.append(line)
    return result

def find_locations(words, text):
    locations = {}
    text = [t.strip(string.punctuation) for t in text]
    for w in words:
        if w.endswith("*"):
            w = w[:-1]
            for i, t in enumerate(text):
                if t.startswith(w): 
                   locations[i] = True
        else:
            for i, t in enumerate(text):
                if t == w: locations[i] = True
    ret = locations.keys()
    ret.sort()
    return ret

def choose_locations(locations, count=2):
    ret = []
    last = 0
    for i in range(count):
        if not locations: break
        l = max(last, locations[0] - 5)
        m = l + 16
        ret.append((l, m))
        last = m
        locations = [l for l in locations if l > last]
    return ret

def handle_search(name, query):
    search = search_text = query.get('search',u'')
    index = data_dir + u'/_swishindex'
    if not search: return handle_sitemap(name, query)

    result = []
    for name in swish(index, search_text):
        if name.startswith("./"): name = name[2:]
        sys.stderr.write("load(%r)\n" % name)
        try:
                text = load(name)
		
        except Error:
                result.append(u'[line ' + quote_markup(name) + ']')
        else:
		entry_meta = {'name': name, 'outer_name': ''}
		entry_text = markup(text, entry_meta)
		entry_plain = html2text(entry_text)
                entry_split = entry_plain.split()

		locations = find_locations(search_text.lower().split(),
                            entry_plain.lower().split())
                extracts = choose_locations(locations)
		summary = entry_meta.get('summary', None)
		title = entry_meta.get('title', name)
		if title != name:
                    result.append(u'[line [link ' +
                                  (quote_markup(title) or u'/') + 
                                  u'[page '+quote_markup(name)+u']]' +
                                  u' \[' + name + u'\]]')
		else:
                    result.append(u'[line [link ' +
                                  (quote_markup(title) or u'/') + 
                                  u'[page '+quote_markup(name)+u']]]')
                if extracts:
                    for a, b in extracts:
                        result.append(u'[line ')
                        if a != 0:
                            result.append("... ")
                        for i in range(a, b):
                            try:
                                e = entry_split[i]
                            except IndexError:
                                continue
                            if i in locations:
                                result.append(" [html <strong>]")
                                result.append(quote_markup(e))
                                result.append("[html </strong>]")
                            else:
                                result.append(" " + quote_markup(e))
                        if b != len(entry_text)-1: 
                            result.append("... ")
                        result.append("]")
		elif summary:
                    result.append(u'[line [html ' + summary + ']]')
		result.append(u'[html <br>]')
			
    if not result:
        result.append('Search found nothing, sorry.')

    result.insert(0, u'[title Search results]\n\n')
    result.insert(1, u'Searching for: [bold ' + quote_markup(search) + u']\n\n')
    result.append(u'[line Search results from swish-e, index updated ' +
                age(index) + ' ago]')
    return make_http_page(string.join(result,u''))