initial commit

1 files changed, 169 insertions, 0 deletions

diff --git a/slack_parser.py b/slack_parser.py
new file mode 100644
index 0000000..8756992
--- /dev/null
+++ b/slack_parser.py
@@ -0,0 +1,169 @@
+import os
+import json
+import re
+import time
+from jsonpath_ng import jsonpath, parse
+
+
+# jsonpath match will match in order of appearence.
+# This mapping defines which position in the RESULT of match to look for which key.
+# so. eg. at 0th position (of the match) will be the key "text", etc.
+key_orderings = {
+    "text": 0,
+    "ts": 1,
+    "name": 2
+}
+
+class SlackParser(object):
+
+    """Docstring for SlackParser. """
+
+    def __init__(self, messagesDir=None, usersFile=None, keys=None):
+
+        self.messagesDir = ""
+        self.usersFile = ""
+        self.keys = list()
+        if messagesDir is not None:
+            self.messagesDir += messagesDir
+        if usersFile is not None:
+            self.usersFile += usersFile
+        if keys is not None:
+            self.keys += keys
+
+        self.usersLib = None
+
+    def getSortedMessageFiles(self):
+        """Docstring for getSortedMessageFiles.
+        Get the files that contain channel data sorted.
+
+        :returns: a sorted list of json files names for the channel
+
+        """
+        files = sorted(os.listdir( self.messagesDir ))
+        return files
+
+    def getUsers(self):
+        """Docstring for getUsers
+        Get the parsed json that contains the raw user info.
+
+        :returns: a dict with all users infos
+
+        """
+        path = self.usersFile
+        # TODO: check for path validity
+        content = json.load(open(path))
+        return content
+
+
+    def buildUsersLibrary(self):
+        """Docstring for buildUsersLibrary
+        Build a Users dict that keeps only usersid to name mappings.
+
+        :returns: a dict user info
+
+        """
+        userLib = {}
+        for userjson in self.getUsers():
+            if "name" in userjson and "id" in userjson:
+                userLib[userjson["id"]] = userjson["name"]
+        self.setUsersLib(userLib)
+        return self
+
+    def setUsersLib(self, usersLib):
+        """Docstring for buildUsersLibrary
+        A setter method that sets the object lib to the given lib.
+
+        """
+        self.usersLib = usersLib
+
+    def getMessages(self, slackFile):
+        """Docstring for getMessages.
+        Given a file, get the messages in that file.
+
+        :slackFile: the json file name containing messages
+        :returns: a list of all messages in the file
+
+        """
+        path = self.messagesDir + "/" + slackFile
+        # TODO: check for path validity
+        content = json.load(open(path))
+        return content
+
+    def getKeyValues(self, slackMessage):
+        """Docstring for getKeyValues.
+        Get the value for each key (defined during init) from the mssg.
+
+        :slackMessage: the parsed json slack message
+        :returns: on success, a dict with only required keys and values
+                  on failure, None
+
+        """
+        data = None
+        # we also check for "subtype" to filter out mssgs like:
+        # "s joined the channel", "x left the channel" etc.
+        if "type" in slackMessage and "subtype" not in slackMessage:
+            if slackMessage["type"] == "message" and slackMessage["text"] != "":
+                data = {}
+                jsonpath_expression = parse("$.." + str(self.keys))
+                # search slackMessage for each key, and add all matches (incl. dups.) in-order
+                match = jsonpath_expression.find(slackMessage)
+                #print(match[0])
+                for key in self.keys:
+                    # TODO: validation for matches possible "index out of range" here.
+                    data[key] = match[key_orderings[key]].value
+                # get the user id in the text
+                userIds = self.getUserIdsFromMessage(data["text"])
+                if userIds != []:
+                    # find corresponding usernames from library
+                    userNamesinMessageDict = self.getUserNamesFromIds(self.usersLib, userIds)
+                    # inject usernames in the text replace userid
+                    data["text"] = self.replaceMessageUserIdsWithNames(userNamesinMessageDict, data["text"])
+                # convert ts to a human readble time
+                data["ts"] = time.strftime( "%H:%M %b %d, %Y", time.localtime(float(data["ts"])) )
+
+        return data
+
+    @staticmethod
+    def getUserIdsFromMessage(text):
+        """Docstring for getUserIdsFromMessage.
+        A helper method that searches a text for Slack user ids. It maps:
+        "<@SLACKID1>adsfasdf <@SLACKID2>" -> ["SLACKID1", "SLACKID2"]
+
+        :text: raw text from Slack message json
+        :returns: a list of userids strings in the text in order of match
+
+        """
+        regex = r"<@(.+?)>"
+        matches = re.findall(regex, text)
+        return matches
+
+    @staticmethod
+    def getUserNamesFromIds(usersDict, userIds):
+        """Docstring for getUserNamesFromIds.
+        A helper method to search a dict for a keys, and get the values.
+
+        :usersDict: a dict library of userid and name mappings
+        :ids: a string list of userids
+        :returns: dict of each userid and its username
+
+        """
+        userMap = {}
+        for userId in userIds:
+            if userId in usersDict:
+                userMap[userId] = usersDict[userId]
+        return userMap
+
+    @staticmethod
+    def replaceMessageUserIdsWithNames(messageUsersDict, text):
+        """Docstring for replaceMessageUserIdsWithNames.
+        A helper method to replace all Slackuser id tags with their @username. So it maps:
+        { "USERID1" : "name1" , "USERID2": "name2" },"<@USERID1> blah blah <@USERID2>." -> "@name1 blah blah @name2."
+
+        :messageUsersDict: the dict containing userid to username mappings for this text
+        :text: the original string contianing Slack userids in it
+        :returns: a string with each Slack userid replaced with mapping value from dict.
+
+        """
+        for userId in messageUsersDict.keys():
+            text = re.sub( "<@{}>".format(userId), "<i style='color: #B18904;'>@{}</i>".format(messageUsersDict[userId]), text )
+        return text