""" emulate a browser with js support using python-spidermonkey
"""

import os
import sys
import traceback
import re
import gc
from lxml import etree

import spidermonkey

here = os.path.dirname(os.path.abspath(__file__))

def get_exception():
    exc, e, tb = sys.exc_info()
    ret = JSPyError(str(exc), str(e), ''.join(traceback.format_tb(tb)))
    del tb
    return ret

def format_stack(stack):
    ret = ''
    for line in stack:
        if '@' not in line:
            ret += line
            continue
        info, fpath = line.rsplit('@', 1)
        fpath = fpath.split('/')[-1]
        if len(info) > 75:
            info = info[:35] + '...' + info[-20:]
        ret += '    %s@%s\n' % (info, fpath)
    return ret


class PyJSError(Exception):
    def __init__(self, name, message, stack):
        self.args = (name, message, stack)
        self.name = name
        self.message = message
        self.stack = stack

    def __str__(self):
        ret = '%s - %s\n' % (self.name, self.message)
        ret += format_stack(self.stack)
        return ret


class JSPyError:
    def __init__(self, name, message, stack):
        self.args = (name, message, stack)
        self.name = name
        self.message = message
        self.stack = stack.split('\n')


class XPathProcessor(object):
    def __init__(self, expr, xml, nodepath, resolvinfo):
        self.expr = expr
        self.xml = xml
        xml = unicode(xml, 'latin-1') # XXX ?!?
        self.resolvinfo = resolvinfo or None
        self.tree = etree.fromstring(xml)
        self.node = self._resolve_xml_path(nodepath)

    def paths(self):
        if self.resolvinfo:
            r = self.node.xpath(self.expr, namespaces=self.resolvinfo)
        else:
            r = self.node.xpath(self.expr)
        return [self._get_path(node) for node in r]

    def _get_path(self, node):
        current = node
        path = []
        if isinstance(node, str) or isinstance(node, unicode):
            if node.is_text:
                path = [0]
                current = current.getparent()
            else:
                parent = node.getparent()
                pparent = parent.getparent()
                parenti = self._get_node_index(parent, parent.getparent())
                path = [parenti + 1]
                current = parent
        while True:
            parent = current.getparent()
            if parent is None:
                break
            i = self._get_node_index(current, parent)
            path.insert(0, i)
            current = parent
        return path

    def _get_node_index(self, node, parent):
        i = 0
        if parent.text:
            i = 1
        for child in parent:
            if child == node:
                break
            if child.tail:
                i += 2
            else:
                i += 1
        else:
            raise AssertionError('node not found in tree!')
        return i

    def _resolve_xml_path(self, path):
        xpexpr = ''.join(['/node()[%s]' % (int(i) + 1,) for
                          i in path.split('/')])
        nodes = self.tree.xpath(xpexpr)
        assert len(nodes) == 1
        return nodes[0]


class XSLTProcessor(object):
    def __init__(self, stylesheet, xml):
        self.stylesheet = stylesheet
        self.xml = xml

    def process(self):
        xsltdoc = etree.fromstring(unicode(self.stylesheet, 'utf-8'))
        xmldoc = etree.fromstring(unicode(self.xml, 'latin-1'))
        open('/tmp/test.xsl', 'w').write(self.stylesheet)
        transform = etree.XSLT(xsltdoc)
        tree = transform(xmldoc)
        return etree.tostring(tree)


class Browser:
    def __init__(self, server, webpath='/'):
        self.server = server
        self.webpath = webpath
        basepath = webpath
        if not basepath.endswith('/'):
            basepath = '/'.join(basepath.split('/')[:-1])
        self.basepath = basepath

    def _init_runtime(self):
        self.runtime = spidermonkey.Runtime(maxbytes=10000000)
        self.context = self.runtime.new_context()

        self.context.bind_class(Browser, bind_constructor=False)
        self.context.bind_class(JSPyError, bind_constructor=True)
        self.context.bind_object('__fakebrowser__', self)

        self._xhrs = {}

    def initialize(self):
        self._init_runtime()
        html = self.server.load_url(self.webpath)
        title = self._get_title(html)
        self.load(os.path.join(here, 'fakebrowser.js'), True)
        self.eval('init_fakebrowser(%r, "http://localhost%s", %r);' % (
                  unicode(html, 'utf-8').encode('utf-8'), self.webpath, title))
        htree = etree.fromstring(html)
        self.resolve_script_tags(htree)
        self.call_onload(htree)
        del htree

    def cleanup(self):
        del self.runtime
        del self.context
        del self._xhrs
        gc.collect()

    def _get_title(self, html):
        match = re.search('<title>(.*)</title>', html, re.M | re.I)
        if match:
            title = match.group(1)
        else:
            title = 'no title'
        return title

    def resolve_script_tags(self, htree):
        scriptnodes = htree.xpath('//script')
        for node in scriptnodes:
            src = node.attrib.get('src')
            if src:
                # XXX first try to load from the server, else from the
                # file system - not sure if that makes sense though...
                try:
                    self.load(src)
                except (KeyError, ValueError, NameError, IOError):
                    self.load(src, True)
            elif node.text:
                self.eval(node.text)

    def call_onload(self, htree):
        body = htree.xpath('//body')[0]
        onload = body.get('onload')
        if onload:
            self.eval(onload)

    def load(self, path, isfs=False):
        """ load and evaluate a JS file

            path is a web-path, either relative from the 'current position'
            (the web page's parent in case this is called directly, the
            JS file's parent in case it's called from a JS file using load())
            or absolute from the web root
        """
        if not isfs:
            if not path.startswith('/'):
                path = self.basepath + path
            js = self.server.load_url(path)
            if js is None:
                # XXX not the right exception imo...
                raise ValueError(
                    'could not load path %s from server' % (path,))
        else:
            if not path.startswith('/'):
                path = os.path.join(here, path)
            fp = open(path)
            try:
                js = fp.read()
            finally:
                fp.close()
        self.eval(js, path)

    def eval(self, js, scriptname='unknown'):
        s = ('''
            try {eval(%r);
            } catch(e) {
                var s = e.toString();
                if (s.indexOf(': ') > -1) {
                    var namemsg = e.toString().split(': ');
                    new JSPyError(namemsg[0], namemsg[1], e.stack || '');
                };
            };
        ''' % (js,))
        try:
            ret = self.context.eval_script(s, scriptname)
        except (SystemExit, KeyboardInterrupt), e:
            raise e
        except:
            exc, e, tb = sys.exc_info()
            print 'an exception occurred in %r' % (scriptname,)
            print '%s - %s' % (exc, e)
            print ''.join(traceback.format_tb(tb))
            raise exc, e, tb
        if isinstance(ret, JSPyError):
            raise PyJSError(ret.name, ret.message, ret.stack)
        return ret

    def send_xhr(self, id, data):
        rdata = {}
        for var in ('method', 'url', 'async', 'user', 'password', 'headers'):
            rdata[var] = self.eval('window.__xhrs__[%s]._%s' % (id, var))
        # async - perform ready state changes and return in the end
        aborted = self.eval('''\
            var xhr = window.__xhrs__[%s];
            for (var i=0; i < 2; i++) {
                xhr._handle_state_change(i+2);
                if (xhr._aborted) {
                    break;
                };
            };
            xhr._aborted;
        ''' % (id,))
        if not aborted:
            ret = self.server.handle_request(rdata['method'], rdata['url'],
                                             rdata['headers'], data)
        return ret

    def process_xpath(self, expr, docxml, nodepath, resolvinfo):
        # print 'processing XPath on xml', docxml[:100], '...'
        ret = XPathProcessor(expr, docxml, nodepath, resolvinfo).paths()
        return ret

    def process_xsl(self, sheet, xml, stack=None):
        # print 'processing XSLT on xml', xml[:100], '...'
        try:
            ret = XSLTProcessor(sheet, xml).process()
        except:
            print 'exception processing XSL, xml: ' + xml
            if stack is not None:
                print 'JavaScript stack trace:'
                print format_stack(stack.split('\n'))
            raise
        return ret

    def raw_input(self, prompt=''):
        return raw_input(prompt)