added more ways to blacklist undesired content

1 files changed, 30 insertions, 15 deletions

diff --git a/src/indeedwatcher.nim b/src/indeedwatcher.nim
index 0fc8af4..a8d6bfc 100644
--- a/src/indeedwatcher.nim
+++ b/src/indeedwatcher.nim
@@ -9,6 +9,14 @@ import sequtils
 import uri
 import json
 
+type
+    indeedJobDesc = object
+        URL: string
+        jobName: string
+        employer: string
+        location: string
+        URLID: string
+
 if not fileExists(getConfigDir() & "/indeedwatcher/config.toml"):
     createDir(getConfigDir() & "/indeedwatcher/")
     let defaultConfig = """
@@ -26,6 +34,8 @@ if not fileExists(getConfigDir() & "/indeedwatcher/config.toml"):
     #Phrases that, if they appear, will cause the job to be instantly thrown out
     [blacklist]
     title= [\"Senior\", \"Sr.\"]
+    employer= [\"NSA\"]
+    location= [\"Ohio\"]
     """
     writeFile(getConfigDir() & "/indeedwatcher/config.toml", defaultConfig)
 
@@ -48,6 +58,8 @@ let cache = open(cachefileloc, fmAppend)
 
 ##Filtering section of config
 let titleblacklist = config["blacklist"]["title"].getElems().mapIt(it.getStr())
+let employerblacklist = config["blacklist"]["employer"].getElems().mapIt(it.getStr())
+let locationblacklist = config["blacklist"]["location"].getElems().mapIt(it.getStr())
 
 ##Does the user desire headlessness?
 var args: JsonNode
@@ -80,6 +92,14 @@ proc terminate() {.noconv.} =
     quit()
 setControlCHook(terminate)
 
+proc postValid(posting: indeedJobDesc) : bool =
+    if any(titleblacklist, proc (input: string): bool = posting.jobName.contains(input)) and
+       any(employerblacklist, proc (input: string): bool = posting.employer.contains(input)) and
+       any(locationblacklist, proc (input: string): bool = posting.location.contains(input)):
+           return true
+    else:
+        return false
+
 for feed in feeds:
     #let args = %*{"desiredCapabilities":{"browserName":"chromium"}}
     session = driver.createSession(args)
@@ -110,22 +130,19 @@ for feed in feeds:
             urlcounter = 0
 
         #Don't even bother visiting it if its in the cache or hits a trigger word
-        var URL = entry.link
-        let queries = URL.parseUri.query.decodeQuery().toSeq()
-        let jobName = queries[0].value
-        let employer = queries[1].value
-        let location = queries[2].value
-        let URLID = queries[3].value
+        let queries = entry.link.parseUri.query.decodeQuery().toSeq()
+        var posting: indeedJobDesc
+        posting = indeedJobDesc(URL: entry.link, jobName: queries[0].value, employer: queries[1].value, location: queries[2].value, URLID: queries[3].value)
 
         #This isn't cache.readFile().contains(URLID)
         #because nim has no way to both open a file in append mode
         #and also open it as reading. Therefore, this blunder, which
         #creates a new file in memory, is used instead.
-        if not readFile(cachefileloc).contains(URLID) and not any(titleblacklist, proc (input: string): bool = jobName.contains(input)):
+        if not readFile(cachefileloc).contains(posting.URLID) and not postValid(posting):
             for attempt in countup(0,3):
                 try:
-                    echo "Telling chromium to navigate to " & URL
-                    session.navigate(URL)
+                    echo "Telling chromium to navigate to " & posting.URL
+                    session.navigate(posting.URL)
                 except:
                     if attempt < 3:
                         echo "Recieved an error: trying again..."
@@ -156,17 +173,15 @@ for feed in feeds:
             URL : $5
             Description:
             $6
-            """ % [jobName, employer, location, salaryInfoAndJobType, URL, fullDesc]
-            writeFile(outdir & jobName.replace("/") & ".txt", output)
+            """ % [posting.jobName, posting.employer, posting.location, salaryInfoAndJobType, posting.URL, fullDesc]
+            writeFile(outdir & posting.jobName.replace("/") & ".txt", output)
             echo "Wrote job to file!"
-            cache.writeLine(URL)
+            cache.writeLine(posting.URL)
             echo "Wrote listing to cache!"
         else:
-            echo URL & " was in cache or hit a trigger word, discarding"
+            echo posting.URL & " was in cache or hit a trigger word, discarding"
     session.close()
 
 
 cache.close()
-#session.close()
 terminate(chromedriver)
-