Merge pull request #347 from spencermountain/dev

Dev

builds/wtf_wikipedia.js

@@ -1,4 +1,4 @@
-/* wtf_wikipedia 8.1.0 MIT */
+/* wtf_wikipedia 8.1.1 MIT */
 (function (global, factory) {
   typeof exports === 'object' && typeof module !== 'undefined' ? module.exports = factory(require('https')) :
   typeof define === 'function' && define.amd ? define(['https'], factory) :
@@ -3893,6 +3893,7 @@
   };
 
   var skipSpanRow = function skipSpanRow(row) {
+    row = row || [];
     var len = row.length;
     var hasTxt = row.filter(function (str) {
       return str;
@@ -3998,7 +3999,15 @@
     .split(/\n/).map(function (l) {
       return l.trim();
     });
-    var rows = _findRows(lines); //remove non-header span rows
+    var rows = _findRows(lines);
+    rows = rows.filter(function (r) {
+      return r;
+    });
+
+    if (rows.length === 0) {
+      return [];
+    } //remove non-header span rows
+
 
     rows = removeMidSpans(rows); //support colspan, rowspan...
 
@@ -8797,7 +8806,7 @@
 
   var category = fetchCategory;
 
-  var _version = '8.1.0';
+  var _version = '8.1.1';
 
   var wtf = function wtf(wiki, options) {
     return _01Document(wiki, options);

builds/wtf_wikipedia.mjs

@@ -1,4 +1,4 @@
-/* wtf_wikipedia 8.1.0 MIT */
+/* wtf_wikipedia 8.1.1 MIT */
 import https from 'https';
 
 var parseUrl = function parseUrl(url) {
@@ -3887,6 +3887,7 @@ var cleanText = function cleanText(str) {
 };
 
 var skipSpanRow = function skipSpanRow(row) {
+  row = row || [];
   var len = row.length;
   var hasTxt = row.filter(function (str) {
     return str;
@@ -3992,7 +3993,15 @@ var parseTable = function parseTable(wiki) {
   .split(/\n/).map(function (l) {
     return l.trim();
   });
-  var rows = _findRows(lines); //remove non-header span rows
+  var rows = _findRows(lines);
+  rows = rows.filter(function (r) {
+    return r;
+  });
+
+  if (rows.length === 0) {
+    return [];
+  } //remove non-header span rows
+
 
   rows = removeMidSpans(rows); //support colspan, rowspan...
 
@@ -8791,7 +8800,7 @@ var fetchCategory = function fetchCategory(category, lang, options) {
 
 var category = fetchCategory;
 
-var _version = '8.1.0';
+var _version = '8.1.1';
 
 var wtf = function wtf(wiki, options) {
   return _01Document(wiki, options);

package.json

@@ -1,7 +1,7 @@
 {
   "name": "wtf_wikipedia",
   "description": "parse wikiscript into json",
-  "version": "8.1.0",
+  "version": "8.1.1",
   "author": "Spencer Kelly <spencermountain@gmail.com> (http://spencermounta.in)",
   "repository": {
     "type": "git",

plugins/classify/src/byInfobox/infoboxes.js

@@ -246,7 +246,7 @@ module.exports = {
   nfl_single_game: 'Event/SportsEvent', //1
   nobility: 'Organization', //5
   non_test_cricket_team: 'Organization/SportsTeam', //4
-  'non-profit': 'SportsTeam', //6
+  'non-profit': 'Organization/SportsTeam', //6
   nrhp: '', //144
   nycs: '', //7
   officeholder: 'Person', //133

plugins/classify/src/index.js

@@ -14,7 +14,7 @@ const plugin = function(models) {
 
     // dont classify these
     if (skipPage(doc, options)) {
-      return null
+      return score(res, options)
     }
 
     //look for 'infobox person', etc

plugins/classify/src/score.js

@@ -63,11 +63,10 @@ const getScore = function(detail) {
       score *= 0.8
     }
   }
-
   return {
     root: root,
     category: category,
-    score: score,
+    score: Math.ceil(score),
     detail: detail
   }
 }

scratch.js

@@ -5,9 +5,10 @@ wtf.extend(require('./plugins/i18n/src'))
 wtf.extend(require('./plugins/summary/src'))
 wtf.extend(require('./plugins/category/src'))
 
-// wtf.fetch('Toronto').then(doc => {
-//   let html = doc.html()
-//   console.log(html)
+// wtf.fetch('Croatian language').then(doc => {
+//   console.log(doc.tables().map(t => t.json()))
+//   // let html = doc.html()
+//   // console.log(html)
 // })
 
 // wtf.fetchCategory('Larchmont, New York').then(res => {
@@ -28,3 +29,7 @@ wtf.extend(require('./plugins/category/src'))
 // let doc = wtf(txt)
 // let res = doc.classify()
 // console.log(res)
+let str = `
+`
+let doc = wtf(str)
+console.log(doc.table())

src/_version.js

@@ -1 +1 @@
-module.exports = '8.1.0'
+module.exports = '8.1.1'

src/table/parse/index.js

@@ -32,6 +32,7 @@ const cleanText = function(str) {
 }
 
 const skipSpanRow = function(row) {
+  row = row || []
   let len = row.length
   let hasTxt = row.filter(str => str).length
   //does it have 3 empty spaces?
@@ -127,6 +128,10 @@ const parseTable = function(wiki) {
     .split(/\n/)
     .map(l => l.trim())
   let rows = findRows(lines)
+  rows = rows.filter(r => r)
+  if (rows.length === 0) {
+    return []
+  }
 
   //remove non-header span rows
   rows = removeMidSpans(rows)