pysearch

pysearch.py

import operator, os, re, urllib
from google.appengine.ext import webapp
from google.appengine.ext.webapp import template

__all__ = ['PySearch']

anchors, lower_anchors = None, None
def get_anchors():
    global anchors, lower_anchors
    if anchors is None:
        anchors, lower_anchors = {}, {}
        genindex = os.path.join(
            os.path.dirname(__file__), 'genindex-all.html')
        genpatt = re.compile(r'<dt><a href="([^"]+)">([^<]*)')
        f = open(genindex)
        for match in re.finditer(genpatt, f.read()):
            ank = match.group(1)
            anchors[ank] = ank
            lower_anchors[ank.lower()] = ank
    return anchors, lower_anchors

def exact_match(q, anchors):
    for tag in ['#%s' % q,         # anchor match (includes builtins)
                '#module-%s' % q,  # module match
                '.%s' % q,         # attribute match
                '-%s' % q,         # something match
               ]:
        for k, v in anchors.items():
            if k.endswith(tag):
                return v

def inner_match(q, anchors):
    for patt in [r'\b%s\b' % q,  # word match
                 r'\b%s' % q,    # start of word match
                 r'%s\b' % q,    # end of word match
                ]:
        reg = re.compile(patt)
        for k, v in anchors.items():
            if re.search(reg, k):
                return v

class PySearch(webapp.RequestHandler):
    def get(self):
        if self.request.get('q'):
            return self.post()
        path = os.path.join(os.path.dirname(__file__), 'pysearch.html')
        self.response.out.write(template.render(path, {}))

    def post(self):
        q = self.request.get('q')
        if not q:
            return self.redirect(request.path)

        # Get our anchor dicts.
        anchors, lower_anchors = get_anchors()

        # Filter down to candidates.
        cands = dict((k, v) for k, v in anchors.items() if q in k)
        if q.islower():
            lower_cands = dict((k, v) for k, v in lower_anchors.items()
                               if q in k)
        else:
            lower_cands = {}

        # Look for an exact match first in both lists.
        for c in [cands, lower_cands]:
            found = exact_match(q, c)
            if found:
                return self.redirect_docs(found)

        # Look for inner matches in both lists.
        for c in [cands, lower_cands]:
            found = inner_match(q, c)
            if found:
                return self.redirect_docs(found)

        # What did you search for? Sheesh
        if c in [cands, lower_cands]:
            found = sorted(c.values())[:1]
            if found:
                return self.redirect_docs(found[0])

        # I give up
        return self.redirect_google(q)

    def redirect_docs(self, path):
        return self.redirect(os.path.join(
            "http://docs.python.org/", path))

    def redirect_google(self, q):
        query = '+'.join(urllib.quote_plus(x) for x in [
            'site:docs.python.org',
            '( inurl:library/ OR inurl:reference/ )',
            q,
        ])
        return self.redirect('http://google.com/search?q=%s' % query)