From: maxious <maxious@lambdacomplex.org>
Date: Sat, 24 Mar 2012 12:01:50 +0000
Subject: Add validation script
X-Git-Url: https://maxious.lambdacomplex.org/git/?p=disclosr.git&a=commitdiff&h=ec5d4d67db0b90be9c18bd541b4434798d559f1d
---
Add validation script


Former-commit-id: 8f0da9f23f1e65c57b3e158ef358630a33f10141
---


--- a/admin/importGov2RegisterRSSFacebookTwitter.php
+++ b/admin/importGov2RegisterRSSFacebookTwitter.php
@@ -1,7 +1,7 @@
 <?php
 
 require_once '../include/common.inc.php';
-require($basePath.'lib/phpquery/phpQuery/phpQuery.php');
+require($basePath . 'lib/phpquery/phpQuery/phpQuery.php');
 
 $db = $server->get_db('disclosr-agencies');
 $rows = $db->get_view("app", "byName")->rows;
@@ -10,41 +10,40 @@
 foreach ($rows as $row) {
     $nametoid[trim($row->key)] = $row->value;
 }
+
 function extractHTMLAccounts($url, $accountType) {
     global $accounts, $nametoid;
     $request = Requests::get($url);
     $doc = phpQuery::newDocumentHTML($request->body);
     phpQuery::selectDocument($doc);
     foreach (pq('tr')->elements as $tr) {
-       //echo $tr->nodeValue.PHP_EOL;
-       $agency = "";
-       $url = "";
-       foreach ($tr->childNodes as $td) {
-           $class = $td->getAttribute("class");
-           //echo "cccc $class ".$td->nodeValue.PHP_EOL;
-           if ($class == "s11" || $class == "s10" || $class == "s7") {
-               $agency = $td->nodeValue;
-           } else if ($class == "s6" || $class == "s9"){
-               $url = $td->nodeValue;
-               foreach($td->childNodes as $a) {
-                   $href = $a->getAttribute("href");
-                   if ($href != "") {
-                       $url = $href;
-                   }
-               }
-           }
-       }
-       if ($agency != "" && $url != "") {
-           if (!in_array(trim($agency), array_keys($nametoid))) {
-                        echo trim($agency)." missing" . PHP_EOL;
-                    } else {
-                     //   echo $agency." = ".$url.PHP_EOL;
-                        $accounts[$nametoid[trim($agency)]][$accountType][] = $url;
+        //echo $tr->nodeValue.PHP_EOL;
+        $agency = "";
+        $url = "";
+        foreach ($tr->childNodes as $td) {
+            $class = $td->getAttribute("class");
+            //echo "cccc $class ".$td->nodeValue.PHP_EOL;
+            if ($class == "s11" || $class == "s10" || $class == "s7") {
+                $agency = $td->nodeValue;
+            } else if ($class == "s6" || $class == "s9") {
+                $url = $td->nodeValue;
+                foreach ($td->childNodes as $a) {
+                    $href = $a->getAttribute("href");
+                    if ($href != "") {
+                        $url = $href;
                     }
-       
-       }
+                }
+            }
+        }
+        if ($agency != "" && $url != "") {
+            if (!in_array(trim($agency), array_keys($nametoid))) {
+                echo trim($agency) . " missing" . PHP_EOL;
+            } else {
+                //   echo $agency." = ".$url.PHP_EOL;
+                $accounts[$nametoid[trim($agency)]][$accountType][] = $url;
+            }
+        }
     }
-    
 }
 
 function extractCSVAccounts($url, $accountType, $nameField, $accountField, $filter) {
@@ -53,7 +52,7 @@
     $Data = str_getcsv($request->body, "\n"); //parse the rows 
     $headers = Array();
     foreach ($Data as $num => $line) {
-        $Row = str_getcsv($line, ",",'"');
+        $Row = str_getcsv($line, ",", '"');
         if ($num == 0) {
             
         } else if ($num == 1) {
@@ -64,7 +63,7 @@
                 $agencyName = $Row[array_search($nameField, $headers)];
                 if (!$filter || $Row[array_search("State", $headers)] == "NAT") {
                     if (!in_array(trim($agencyName), array_keys($nametoid))) {
-                        echo trim($agencyName)." missing" . PHP_EOL;
+                        echo trim($agencyName) . " missing" . PHP_EOL;
                     } else {
                         // echo $Row[array_search($nameField, $headers)] . PHP_EOL;
                         $accounts[$nametoid[trim($agencyName)]][$accountType][] = $Row[array_search($accountField, $headers)];
@@ -84,6 +83,18 @@
 extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGJxandJREhLSGlWWUZfZ2xKOTNHZ0E&output=html", "RSS");
 // facebook 
 extractHTMLAccounts("https://docs.google.com/spreadsheet/pub?hl=en_GB&hl=en_GB&key=0Ah41IAK0HzSTdGtjcW9vOXdyZ3pOV21vQU51VmhzQnc&single=true&gid=0&output=html", "Facebook");
+foreach ($accounts as $id => $accountTypes) {
+    echo $id . "<br>" . PHP_EOL;
+    $doc = object_to_array($db->get($id));
+    // print_r($doc);
 
+    foreach ($accountTypes as $accountType => $accounts) {
+        if (!isset($doc["has" . $accountType]) || !is_array($doc["has" . $accountType])) {
+            $doc["has" . $accountType] = Array();
+        }
+        $doc["has" . $accountType] = array_unique(array_merge($doc["has" . $accountType], $accounts));
+    }
+    $db->save($doc);
+}
 ?>
 

--- /dev/null
+++ b/admin/validation.py
@@ -1,1 +1,30 @@
+#http://packages.python.org/CouchDB/client.html
+import couchdb
+import json
+import pprint
+import re
+from tidylib import tidy_document
 
+couch = couchdb.Server('http://127.0.0.1:5984/')
+
+# select database
+docsdb = couch['disclosr-documents']
+
+def f(x):
+	invalid = re.compile(r"ensure|testing|flicker|updating|longdesc|Accessibility Checks|not recognized")
+	valid = re.compile(r"line")
+	return (not invalid.search(x)) and valid.search(x) and x != ''
+
+for row in docsdb.view('app/getValidationRequired'): 
+    print row.id
+    html = docsdb.get_attachment(row.id,row.value.iterkeys().next()).read()
+    #print html
+    document, errors = tidy_document(html,options={'accessibility-check':1,'show-warnings':0,'markup':0},keep_doc=True)
+    #http://www.aprompt.ca/Tidy/accessibilitychecks.html
+    #print document
+    errors = '\n'.join(filter(f,errors.split('\n')))
+    #print errors
+    doc = docsdb.get(row.id)
+    doc['validation'] = errors
+    docsdb.save(doc)
+

--- a/couchdb/settee
+++ b/couchdb/settee

--- a/include/couchdb.inc.php
+++ b/include/couchdb.inc.php
@@ -22,7 +22,10 @@
        },
        "byWebServer": {
            "map": "function(doc) {\n  emit(doc.web_server, doc);\n}"
-       }
+       },
+  "getValidationRequired": {
+       "map": "function(doc) {\nif (doc.mime_type == \"text/html\" \n&& typeof(doc.validation) == \"undefined\") {\n  emit(doc._id, doc._attachments);\n}\n}"
+   }
    }*/
 }
 
@@ -167,3 +170,4 @@
 function setteErrorHandler($e) {
     echo $e->getMessage() . "<br>" . PHP_EOL;
 }
+