# -*- coding: utf-8 -*-
from ctypes import *

# Copyright Michael Anthony Puls II, http://shadow2531.com/
# Distributed under the Boost Software License, Version 1.0.
# http://boost.org/LICENSE_1_0.txt .
# Newest versions:  http://shadow2531.com/opera/testcases/mailto/MailtoURIParserPack.zip

# This is a python (more or less a direct convert) version of the c++ MailtoURIParser at http://shadow2531.com/cpp/mailto_funcs11.zip
# It also uses the exact same test cases.

# Make sure all newline pairs or single newlines are represented in the string as \n
def normalizeNewlines(s):
    return s.replace("\r\n", "\n").replace("\r", "\n")

# Decode %HH. Treat invalid %HH literally.
# The result is a string of raw UTF-8 sequences
def safe_decode(s):
    hexits = "0123456789ABCDEF"
    ret = ""
    i = 0
    size = len(s)
    while i < size:
        c = s[i]
        if c == "%" and i + 2 < size:
            h1 = s[i + 1].upper()
            h2 = s[i + 2].upper()
            a = hexits.find(h1)
            b = hexits.find(h2)
            if a == -1 or b == -1:
                ret += c
            else:
                x = "%c" % (a * 16 + b)
                ret += x
                i += 2
        else:
            ret += c
        i += 1
    return normalizeNewlines(ret)

# encode hvalue data
def encodex(s):
    hexits = "0123456789ABCDEF"
    noencode = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-_.!*()"
    ret = ""
    for i in s:
        if i == '\n':
            ret += "%0D%0A"
        elif noencode.find(i) != -1:
            ret += i
        else:
            ret += '%'
            i_as_char = c_char(i)
            unsigned_i_as_char = cast(i_as_char.value, POINTER(c_ubyte));
            h1 = hexits[unsigned_i_as_char.contents.value >> 4]
            h2 = hexits[unsigned_i_as_char.contents.value & 0x0F]
            ret += h1
            ret += h2
    return ret

class MailtoURIParser:
    def __init__(self, inuri):
        self.__eval(inuri)
    def setURI(self, inuri):
        self.__eval(inuri)
    def getURI(self):
        return self.__uri
    def getNormalizedURI(self):
        return self.__nuri
    def getEncodedTO(self):
        return self.__to
    def getDecodedTO(self):
        return self.__dto
    def getEncodedSubject(self):
        return self.__subject
    def getDecodedSubject(self):
        return self.__dsubject
    def getEncodedBody(self):
        return self.__body
    def getDecodedBody(self):
        return self.__dbody
    def getEncodedCC(self):
        return self.__cc
    def getDecodedCC(self):
        return self.__dcc
    def getEncodedBCC(self):
        return self.__bcc
    def getDecodedBCC(self):
        return self.__dbcc

    # %w = original uri
    # %n = normalized uri (mailto URI with just the basic hnames with no duplicate hnames)
    # %t = decoded TO value
    # %T = encoded TO value
    # %s = decoded Subject value
    # %m = deocded Body value
    # %M = encoded Body value
    # %c = decoded CC value
    # %C = encoded CC value
    # %b = decoded BCC value
    # %B = encoded BCC value
    # %% = %
    # An invalid %key or a % at the end of the string is treated literally.

    # By example:
    # resolveCommandFormatString("%T") would return the value of the to string.
    # resolveCommandFormatString("%%") would return %.

    def resolveCommandFormatString(self, s):
        ret = ""
        ident = "%"
        i = 0
        size = len(s)
        while i < size:
            c = s[i]
            if c == ident and i + 1 < size:
                next = s[i + 1]
                if next == ident:
                    ret += ident
                elif next == "w":
                    ret += self.__uri
                elif next == "n":
                    ret += self.__nuri
                elif next == "T":
                    ret += self.__to
                elif next == "t":
                    ret += self.__dto
                elif next == "S":
                    ret += self.__subject
                elif next == "s":
                    ret += self.__dsubject
                elif next == "M":
                    ret += self.__body
                elif next == "m":
                    ret += self.__dbody
                elif next == "C":
                    ret += self.__cc
                elif next == "c":
                    ret += self.__dcc
                elif next == "B":
                    ret += self.__bcc
                elif next == "b":
                    ret += self.__dbcc
                else:
                    ret += c
                    ret += next
                i += 1
            else:
                ret += c
            i += 1
        return ret

    # Private variables
    __uri = __nuri = __to = __dto = __subject = __dsubject = __body = __dbody = __cc = __dcc = __bcc = __dbcc = ""

    # Split the parsable data and store it. Rules are from http://shadow2531.com/opera/testcases/mailto/rfc2368-3.html
    # For to, cc and bcc: Join all non-empty hvalues by %2C%20 and decode
    # For body: Join the first non-empty hvalue and all hvalues (even if they're empty) after that with %0D%0A and decode
    # For subject: Use only the last subject hvalue (even if it's empty and even if a previous one is not) and decode

    def __storeData(self, sdata):
        self.__to = self.__subject = self.__body = self.__cc = self.__bcc = ""
        hlist = sdata.split('&')
        for i in hlist:
            eq = i.find('=')
            if eq == -1:
                continue
            hname = i[:eq].lower()
            if hname == "to":
                value = i[eq + 1:]
                if value != "":
                    if self.__to != "":
                        self.__to += "%2C%20"
                    self.__to += value
            elif hname == "cc":
                value = i[eq + 1:]
                if value != "":
                    if self.__cc != "":
                        self.__cc += "%2C%20"
                    self.__cc += value
            elif hname == "bcc":
                value = i[eq + 1:]
                if value != "":
                    if self.__bcc != "":
                        self.__bcc += "%2C%20"
                    self.__bcc += value
            elif hname == "subject":
                self.__subject = i[eq + 1:]
            elif hname == "body":
                value = i[eq + 1:]
                if (self.__body == "" and value == "") == 0:
                    if self.__body != "":
                        self.__body += "%0D%0A"
                    self.__body += value

        self.__dto = safe_decode(self.__to);
        self.__dcc = safe_decode(self.__cc);
        self.__dbcc = safe_decode(self.__bcc);
        self.__dsubject = safe_decode(self.__subject);
        self.__dbody = safe_decode(self.__body);

        # Recreate the encoded values from the decoded ones to fix any chars that should not be encoded or fix ones that should, but are not.

        self.__to = encodex(self.__dto);
        self.__subject = encodex(self.__dsubject);
        self.__body = encodex(self.__dbody);
        self.__cc = encodex(self.__dcc);
        self.__bcc = encodex(self.__dbcc)

    # Whenever a URI is set, if it's a mailto URI, parse and store the data.
    # If it's not a mailto URI, reset the data.  setURI("") can be used to
    # clear the data.

    def __eval(self, inuri):
        if inuri.lower().startswith("mailto:"):
            parseable = inuri[4:]
            parseable = parseable.replace(':', '=', 1)
            parseable = parseable.replace('?', '&', 1)
            self.__storeData(parseable)
            self.__uri = inuri

            # Create a normalized mailto URI with just the basic hnames and no duplicate hnames (for mail clients that can't handle duplicate hnames)
            self.__nuri = "mailto:";
            self.__nuri += self.__to;
            self.__nuri += "?subject=";
            self.__nuri += self.__subject;
            self.__nuri += "&body=";
            self.__nuri += self.__body;
            self.__nuri += "&cc=";
            self.__nuri += self.__cc;
            self.__nuri += "&bcc=";
            self.__nuri += self.__bcc;
        else:
            self.__uri = self.__nuri = "mailto:"
            self.__to = self.__dto = self.__subject = self.__dsubject = self.__body = self.__dbody = self.__cc = self.__dcc = self.__bcc = self.__dbcc = ""

def test(t, inc, out):
    if inc == out:
        print "test", t, "= pass"
    else:
        print "test", t, "= fail"
        print "before_" + inc + "_after"

# Test cases to show usage

x = MailtoURIParser("")
test(1, x.getEncodedTO(), "")
x.setURI("mailto:")
test(2, x.getEncodedTO(), "")
x.setURI("mailto:?");
test(3, x.getEncodedTO(), "")
x.setURI("mailto:email%40site.com")
test(4, x.getEncodedTO(), "email%40site.com")
x.setURI("mailto:?to=email%40site.com")
test(5, x.getEncodedTO(), "email%40site.com")
x.setURI("mailto:email1%40site.com%2C%20email2%40site.com?to=email3%40site.com%2C%20email4%40site.com")
test(6, x.getEncodedTO(), "email1%40site.com%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com")
x.setURI("mailto:?to=email1%40site.com&to=email2%40site.com&to=email3%40site.com")
test(7, x.getEncodedTO(), "email1%40site.com%2C%20email2%40site.com%2C%20email3%40site.com")
x.setURI("mailto:?to=&to=&to=&to=&to=&to=&to=&to=&to=&to=&to=&to=&to=")
test(8, x.getEncodedTO(), "")
x.setURI("mailto:?to=&to=email1%40site.com&to=&to=email2%40site.com&to=")
test(9, x.getEncodedTO(), "email1%40site.com%2C%20email2%40site.com")
x.setURI("mailto:?subject=bark%20bark")
test(10, x.getEncodedSubject(), "bark%20bark")
x.setURI("mailto:?subject=")
test(11, x.getEncodedSubject(), "")
x.setURI("mailto:?subject=&subject=")
test(12, x.getEncodedSubject(), "")
x.setURI("mailto:?subject=1&subject=2&subject=last%20one")
test(13, x.getEncodedSubject(), "last%20one")
x.setURI("mailto:?subject=1&subject=")
test(14, x.getEncodedSubject(), "")
x.setURI("mailto:?body=line1%0D%0Aline2")
test(15, x.getEncodedBody(), "line1%0D%0Aline2")
x.setURI("mailto:?body=line1&body=line2")
test(16, x.getEncodedBody(), "line1%0D%0Aline2")
x.setURI("mailto:?body=&body=&body=line1")
test(17, x.getEncodedBody(), "line1")
x.setURI("mailto:?body=&body=&body=line1&body=&body=&body=line4&body=&body=")
test(18, x.getEncodedBody(), "line1%0D%0A%0D%0A%0D%0Aline4%0D%0A%0D%0A")
x.setURI("mailto:?body=line1&body=")
test(19, x.getEncodedBody(), "line1%0D%0A")
x.setURI("mailto:email1%40site.com?to=&to=email2%40site.com&subject=1&subject=&subject=2&subject=&body=&body=&body=line1&body=&body=&body=line4&body=&cc=zam%40site.com&bcc=bam%40site.com")
test(20, x.getNormalizedURI(), "mailto:email1%40site.com%2C%20email2%40site.com?subject=&body=line1%0D%0A%0D%0A%0D%0Aline4%0D%0A&cc=zam%40site.com&bcc=bam%40site.com")
big = "mailto:%22a%5C%5Cb%22%20%3Cemail1%40site.com%3E%2C%20email2%40site.com?to=&to=email3%40site.com%2C%20email4%40site.com&to=email5%40site.com%2C%20email6%40site.com&to=&to=&subject=You%20should%20not%20see%20me%2E&subject=last%20call%20for%20cats%20%26%20dogs%2E&body=&body=&body=&body=line1%0D%0Aline2&body=&body=&body=&body=line6&body=line7%0D%0A1+2+3+4+5%0D%0AColumn1A%09Column1B%09Column1C%09Column1D%0D%0A%E2%88%9A&cc=%22a%5C%5Cb%22%20%3Csomeone1%40site.com%3E%2C%20someone2%40site.com&cc=&cc=someone3%40site.com%2C%20someone4%40site.com%2C%20%22foo%20%5C%22bar%5C%22%22%20%3Cfoo%40bar%2Ecom%3E&cc=&cc=&bcc=%22a%5C%5Cb%22%20%3Csomeoneelse1%40site.com%3E%2C%20someoneelse2%40site.com&bcc=&bcc=someoneelse3%40site.com%2C%20someoneelse4%40site.com&bcc=&bcc="
x.setURI(big)
test(21, x.getNormalizedURI(), "mailto:%22a%5C%5Cb%22%20%3Cemail1%40site.com%3E%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com%2C%20email5%40site.com%2C%20email6%40site.com?subject=last%20call%20for%20cats%20%26%20dogs.&body=line1%0D%0Aline2%0D%0A%0D%0A%0D%0A%0D%0Aline6%0D%0Aline7%0D%0A1%2B2%2B3%2B4%2B5%0D%0AColumn1A%09Column1B%09Column1C%09Column1D%0D%0A%E2%88%9A&cc=%22a%5C%5Cb%22%20%3Csomeone1%40site.com%3E%2C%20someone2%40site.com%2C%20someone3%40site.com%2C%20someone4%40site.com%2C%20%22foo%20%5C%22bar%5C%22%22%20%3Cfoo%40bar.com%3E&bcc=%22a%5C%5Cb%22%20%3Csomeoneelse1%40site.com%3E%2C%20someoneelse2%40site.com%2C%20someoneelse3%40site.com%2C%20someoneelse4%40site.com")
test(22, x.resolveCommandFormatString("\"c:\\program files\\program\\program.exe\" \"%T\" \"%S\" \"%M\" \"%C\" \"%B\""), "\"c:\\program files\\program\\program.exe\" \"%22a%5C%5Cb%22%20%3Cemail1%40site.com%3E%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com%2C%20email5%40site.com%2C%20email6%40site.com\" \"last%20call%20for%20cats%20%26%20dogs.\" \"line1%0D%0Aline2%0D%0A%0D%0A%0D%0A%0D%0Aline6%0D%0Aline7%0D%0A1%2B2%2B3%2B4%2B5%0D%0AColumn1A%09Column1B%09Column1C%09Column1D%0D%0A%E2%88%9A\" \"%22a%5C%5Cb%22%20%3Csomeone1%40site.com%3E%2C%20someone2%40site.com%2C%20someone3%40site.com%2C%20someone4%40site.com%2C%20%22foo%20%5C%22bar%5C%22%22%20%3Cfoo%40bar.com%3E\" \"%22a%5C%5Cb%22%20%3Csomeoneelse1%40site.com%3E%2C%20someoneelse2%40site.com%2C%20someoneelse3%40site.com%2C%20someoneelse4%40site.com\"")
test(23, x.resolveCommandFormatString("%%T"), "%T")
x.setURI("mailto:?subject=%E2%88%9A")
test(24, x.getDecodedSubject(), "\xE2\x88\x9A")
x.setURI("mailto:?subject=1+2+3+4+5")
test(25, x.getDecodedSubject(), "1+2+3+4+5")
x.setURI("mailto:?subject=%G3")
test(26, x.getDecodedSubject(), "%G3")
x.setURI("mailto:?subject=%%%")
test(27, x.getDecodedSubject(), "%%%")
x.setURI("mailto:?subject=%")
test(28, x.getDecodedSubject(), "%")
x.setURI("mailto:?body=%0A%0D%0A%0D")
test(29, x.getDecodedBody(), "\n\n\n")
x.setURI("mailto:?&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&body=line1")
test(30, x.getEncodedBody(), "line1")
x.setURI("")
test(31, x.getEncodedCC(), "")
x.setURI("mailto:")
test(32, x.getEncodedCC(), "")
x.setURI("mailto:?")
test(33, x.getEncodedCC(), "")
x.setURI("mailto:?cc=email%40site.com")
test(34, x.getEncodedCC(), "email%40site.com")
x.setURI("mailto:?cc=email1%40site.com%2C%20email2%40site.com&cc=email3%40site.com%2C%20email4%40site.com")
test(35, x.getEncodedCC(), "email1%40site.com%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com")
x.setURI("mailto:?cc=email1%40site.com&cc=email2%40site.com&cc=email3%40site.com")
test(36, x.getEncodedCC(), "email1%40site.com%2C%20email2%40site.com%2C%20email3%40site.com")
x.setURI("mailto:?cc=&cc=&cc=&cc=&cc=&cc=&cc=&cc=&cc=&cc=&cc=&cc=&cc=")
test(37, x.getEncodedCC(), "")
x.setURI("mailto:?cc=&cc=email1%40site.com&cc=&cc=email2%40site.com&cc=")
test(38, x.getEncodedCC(), "email1%40site.com%2C%20email2%40site.com")
x.setURI("")
test(39, x.getEncodedBCC(), "")
x.setURI("mailto:")
test(40, x.getEncodedBCC(), "")
x.setURI("mailto:?")
test(41, x.getEncodedBCC(), "")
x.setURI("mailto:?bcc=email%40site.com")
test(42, x.getEncodedBCC(), "email%40site.com")
x.setURI("mailto:?bcc=email1%40site.com%2C%20email2%40site.com&bcc=email3%40site.com%2C%20email4%40site.com")
test(43, x.getEncodedBCC(), "email1%40site.com%2C%20email2%40site.com%2C%20email3%40site.com%2C%20email4%40site.com")
x.setURI("mailto:?bcc=email1%40site.com&bcc=email2%40site.com&bcc=email3%40site.com")
test(44, x.getEncodedBCC(), "email1%40site.com%2C%20email2%40site.com%2C%20email3%40site.com")
x.setURI("mailto:?bcc=&bcc=&bcc=&bcc=&bcc=&bcc=&bcc=&bcc=&bcc=&bcc=&bcc=&bcc=&bcc=")
test(45, x.getEncodedBCC(), "")
x.setURI("mailto:?bcc=&bcc=email1%40site.com&bcc=&bcc=email2%40site.com&bcc=")
test(46, x.getEncodedBCC(), "email1%40site.com%2C%20email2%40site.com")
x.setURI("mailto:?subject=m%26m%09bob")
test(47, x.getDecodedSubject(), "m&m\tbob")
x.setURI("mailto:email%40site.com")
test(48, x.resolveCommandFormatString("%%%%%t"), "%%email@site.com")
x.setURI("mailto:?cc=email%40site.com")
test(49, x.getDecodedCC(), "email@site.com")
x.setURI("mailto:?to=email%40site.com")
test(50, x.getDecodedTO(), "email@site.com")
x.setURI("MaIlTo:?BcC=email%40site.com")
test(51, x.getDecodedBCC(), "email@site.com")
x.setURI(big)
fmt = "program_that_wants_decoded_values -to \"%t\" -subj \"%s\" -cc \"%c\" -bcc \"%b\" -body \"%m\""
test(52, x.resolveCommandFormatString(fmt), "program_that_wants_decoded_values -to \"\"a\\\\b\" <email1@site.com>, email2@site.com, email3@site.com, email4@site.com, email5@site.com, email6@site.com\" -subj \"last call for cats & dogs.\" -cc \"\"a\\\\b\" <someone1@site.com>, someone2@site.com, someone3@site.com, someone4@site.com, \"foo \\\"bar\\\"\" <foo@bar.com>\" -bcc \"\"a\\\\b\" <someoneelse1@site.com>, someoneelse2@site.com, someoneelse3@site.com, someoneelse4@site.com\" -body \"line1\nline2\n\n\n\nline6\nline7\n1+2+3+4+5\nColumn1A\tColumn1B\tColumn1C\tColumn1D\n\xE2\x88\x9A\"")
x.setURI("")
test(53, x.getEncodedTO(), "")
test(54, x.getEncodedBCC(), "")
test(55, x.getEncodedCC(), "")
test(56, x.getEncodedBody(), "")
test(57, x.getEncodedSubject(), "")
test(58, x.getURI(), "mailto:")
test(59, x.getNormalizedURI(), "mailto:")
test(60, x.getDecodedTO(), "")
test(61, x.getDecodedBCC(), "")
test(62, x.getDecodedCC(), "")
test(63, x.getDecodedBody(), "")
test(64, x.getDecodedSubject(), "")
x.setURI("mailto:?body=+++")
test(65, x.getEncodedBody(), "%2B%2B%2B")
test(66, x.getDecodedBody(), "+++")