From db4cd06576a783bd2daef567141c9f4b3a2bf70c Mon Sep 17 00:00:00 2001
From: Julian Simioni <julian@simioni.org>
Date: Mon, 29 Oct 2018 13:03:07 -0400
Subject: [PATCH] feat(log): Add full clean context to parser logs

Without the unmodified text input, it's hard to tell what cases is
causing these messages.
---
 query/autocomplete.js          |  2 +-
 query/search_original.js       |  2 +-
 query/text_parser_addressit.js | 44 +++++++++++++++++-----------------
 sanitizer/_text_addressit.js   |  8 +++----
 4 files changed, 27 insertions(+), 29 deletions(-)

diff --git a/query/autocomplete.js b/query/autocomplete.js
index f16adc44c..0f084d33d 100644
--- a/query/autocomplete.js
+++ b/query/autocomplete.js
@@ -127,7 +127,7 @@ function generateQuery( clean ){
 
   // run the address parser
   if( clean.parsed_text ){
-    textParser( clean.parsed_text, vs );
+    textParser( clean, vs );
   }
 
   return {
diff --git a/query/search_original.js b/query/search_original.js
index 2bcaeb700..09f671e06 100644
--- a/query/search_original.js
+++ b/query/search_original.js
@@ -130,7 +130,7 @@ function generateQuery( clean ){
 
   // run the address parser
   if( clean.parsed_text ){
-    textParser( clean.parsed_text, vs );
+    textParser( clean, vs );
   }
 
   return {
diff --git a/query/text_parser_addressit.js b/query/text_parser_addressit.js
index 72a82831a..15a77aa1b 100644
--- a/query/text_parser_addressit.js
+++ b/query/text_parser_addressit.js
@@ -15,58 +15,58 @@ var adminFields = placeTypes.concat([
 **/
 
 // all the address parsing logic
-function addParsedVariablesToQueryVariables( parsed_text, vs ){
+function addParsedVariablesToQueryVariables( clean, vs ){
 
   // is it a street address?
-  var isStreetAddress = parsed_text.hasOwnProperty('number') && parsed_text.hasOwnProperty('street');
+  var isStreetAddress = clean.parsed_text.hasOwnProperty('number') && clean.parsed_text.hasOwnProperty('street');
   if( isStreetAddress ){
-    vs.var( 'input:name', parsed_text.number + ' ' + parsed_text.street );
+    vs.var( 'input:name', clean.parsed_text.number + ' ' + clean.parsed_text.street );
   }
 
   // ?
-  else if( parsed_text.admin_parts ) {
-    vs.var( 'input:name', parsed_text.name );
+  else if( clean.parsed_text.admin_parts ) {
+    vs.var( 'input:name', clean.parsed_text.name );
   }
 
   // ?
   else {
     logger.warn( 'chaos monkey asks: what happens now?', {
-      parsed_text: parsed_text
+      params: clean
     });
   }
 
   // ==== add parsed matches [address components] ====
 
   // house number
-  if( parsed_text.hasOwnProperty('number') ){
-    vs.var( 'input:housenumber', parsed_text.number );
+  if( clean.parsed_text.hasOwnProperty('number') ){
+    vs.var( 'input:housenumber', clean.parsed_text.number );
   }
 
   // street name
-  if( parsed_text.hasOwnProperty('street') ){
-    vs.var( 'input:street', parsed_text.street );
+  if( clean.parsed_text.hasOwnProperty('street') ){
+    vs.var( 'input:street', clean.parsed_text.street );
   }
 
   // postal code
-  if( parsed_text.hasOwnProperty('postalcode') ){
-    vs.var( 'input:postcode', parsed_text.postalcode );
+  if( clean.parsed_text.hasOwnProperty('postalcode') ){
+    vs.var( 'input:postcode', clean.parsed_text.postalcode );
   }
 
   // ==== add parsed matches [admin components] ====
 
   // city
-  if( parsed_text.hasOwnProperty('city') ){
-    vs.var( 'input:county', parsed_text.city );
+  if( clean.parsed_text.hasOwnProperty('city') ){
+    vs.var( 'input:county', clean.parsed_text.city );
   }
 
   // state
-  if( parsed_text.hasOwnProperty('state') ){
-    vs.var( 'input:region_a', parsed_text.state );
+  if( clean.parsed_text.hasOwnProperty('state') ){
+    vs.var( 'input:region_a', clean.parsed_text.state );
   }
 
   // country
-  if( parsed_text.hasOwnProperty('country') ){
-    vs.var( 'input:country_a', parsed_text.country );
+  if( clean.parsed_text.hasOwnProperty('country') ){
+    vs.var( 'input:country_a', clean.parsed_text.country );
   }
 
   // ==== deal with the 'leftover' components ====
@@ -74,11 +74,11 @@ function addParsedVariablesToQueryVariables( parsed_text, vs ){
 
   // a concept called 'leftovers' which is just 'admin_parts' /or 'regions'.
   var leftoversString = '';
-  if( parsed_text.hasOwnProperty('admin_parts') ){
-    leftoversString = parsed_text.admin_parts;
+  if( clean.parsed_text.hasOwnProperty('admin_parts') ){
+    leftoversString = clean.parsed_text.admin_parts;
   }
-  else if( parsed_text.hasOwnProperty('regions') ){
-    leftoversString = parsed_text.regions.join(' ');
+  else if( clean.parsed_text.hasOwnProperty('regions') ){
+    leftoversString = clean.parsed_text.regions.join(' ');
   }
 
   // if we have 'leftovers' then assign them to any fields which
diff --git a/sanitizer/_text_addressit.js b/sanitizer/_text_addressit.js
index 70a4bd24a..1918308eb 100644
--- a/sanitizer/_text_addressit.js
+++ b/sanitizer/_text_addressit.js
@@ -29,7 +29,7 @@ function _sanitize( raw, clean ){
     delete clean.parsed_text;
 
     // parse text with query parser
-    var parsed_text = parse(clean.text);
+    var parsed_text = parse(clean.text, clean);
     if (check.assigned(parsed_text)) {
       clean.parsed_text = parsed_text;
     }
@@ -51,7 +51,7 @@ module.exports = () => ({
 // this is the addressit functionality from https://github.com/pelias/text-analyzer/blob/master/src/addressItParser.js
 var DELIM = ',';
 
-function parse(query) {
+function parse(query, clean) {
   var getAdminPartsBySplittingOnDelim = function(queryParts) {
     // naive approach - for admin matching during query time
     // split 'flatiron, new york, ny' into 'flatiron' and 'new york, ny'
@@ -110,12 +110,10 @@ function parse(query) {
   // if all we found was regions, ignore it as it is not enough information to make smarter decisions
   if (Object.keys(parsed_text).length === 1 && !_.isUndefined(parsed_text.regions)) {
     logger.info('Ignoring address parser output, regions only', {
-      text: query.text,
-      parsed: parsed_text
+      params: clean
     });
     return null;
   }
 
   return parsed_text;
-
 }