String frequency support (#1989)

* Restore support for string frequency values

* Add support for {value, displayValue} frequencies

* Update test data

* Improve number parsing of string frequencies

* Improve reading detection

* Expose a displayValue property for frequency information

* Update docs

* Expose displayValue to Anki note data

* Fix translator

* Update display generation

* Update test data

* Update counts

7 files changed, 147 insertions, 42 deletions

diff --git a/ext/data/schemas/dictionary-kanji-meta-bank-v3-schema.json b/ext/data/schemas/dictionary-kanji-meta-bank-v3-schema.json
index 49f7c813..0864c9aa 100644
--- a/ext/data/schemas/dictionary-kanji-meta-bank-v3-schema.json
+++ b/ext/data/schemas/dictionary-kanji-meta-bank-v3-schema.json
@@ -1,5 +1,29 @@
 {
     "$schema": "http://json-schema.org/draft-07/schema#",
+    "definitions": {
+        "frequency": {
+            "oneOf": [
+                {
+                    "type": ["string", "number"]
+                },
+                {
+                    "type": "object",
+                    "additionalProperties": false,
+                    "required": [
+                        "value"
+                    ],
+                    "properties": {
+                        "value": {
+                            "type": "number"
+                        },
+                        "displayValue": {
+                            "type": "string"
+                        }
+                    }
+                }
+            ]
+        }
+    },
     "type": "array",
     "description": "Custom metadata for kanji characters.",
     "additionalItems": {
@@ -17,7 +41,7 @@
                 "description": "Type of data. \"freq\" corresponds to frequency information."
             },
             {
-                "type": ["number"],
+                "$ref": "#/definitions/frequency",
                 "description": "Data for the character."
             }
         ]
diff --git a/ext/data/schemas/dictionary-term-meta-bank-v3-schema.json b/ext/data/schemas/dictionary-term-meta-bank-v3-schema.json
index 206e7152..96f2e54b 100644
--- a/ext/data/schemas/dictionary-term-meta-bank-v3-schema.json
+++ b/ext/data/schemas/dictionary-term-meta-bank-v3-schema.json
@@ -1,5 +1,29 @@
 {
     "$schema": "http://json-schema.org/draft-07/schema#",
+    "definitions": {
+        "frequency": {
+            "oneOf": [
+                {
+                    "type": ["string", "number"]
+                },
+                {
+                    "type": "object",
+                    "additionalProperties": false,
+                    "required": [
+                        "value"
+                    ],
+                    "properties": {
+                        "value": {
+                            "type": "number"
+                        },
+                        "displayValue": {
+                            "type": "string"
+                        }
+                    }
+                }
+            ]
+        }
+    },
     "type": "array",
     "description": "Custom metadata for terms.",
     "additionalItems": {
@@ -28,7 +52,7 @@
                     {
                         "oneOf": [
                             {
-                                "type": ["number"],
+                                "$ref": "#/definitions/frequency",
                                 "description": "Frequency information for the term."
                             },
                             {
@@ -44,7 +68,7 @@
                                         "description": "Reading for the term."
                                     },
                                     "frequency": {
-                                        "type": ["number"],
+                                        "$ref": "#/definitions/frequency",
                                         "description": "Frequency information for the term."
                                     }
                                 }
diff --git a/ext/display-templates.html b/ext/display-templates.html
index 534ad704..c181a64e 100644
--- a/ext/display-templates.html
+++ b/ext/display-templates.html
@@ -71,13 +71,13 @@
             <rt class="frequency-disambiguation-reading"></rt>
         </ruby></span>
         <span class="frequency-separator"></span>
-        <span class="frequency-value"></span>
+        <span class="frequency-value-list"></span>
     </span></span>
 </span></span></template>
 <template id="kanji-frequency-item-template" data-remove-whitespace-text="true"><span class="frequency-item"><span class="tag tag-has-body frequency-tag" data-category="frequency" data-frequency-type="kanji">
     <span class="tag-label"><span class="tag-label-content"></span></span>
     <span class="tag-body"><span class="tag-body-content frequency-body">
-        <span class="frequency-value"></span>
+        <span class="frequency-value-list"></span>
     </span></span>
 </span></span></template>
 
diff --git a/ext/js/data/sandbox/anki-note-data-creator.js b/ext/js/data/sandbox/anki-note-data-creator.js
index b7abc0c5..8d363134 100644
--- a/ext/js/data/sandbox/anki-note-data-creator.js
+++ b/ext/js/data/sandbox/anki-note-data-creator.js
@@ -243,7 +243,7 @@ class AnkiNoteDataCreator {
 
     _getKanjiFrequencies(dictionaryEntry) {
         const results = [];
-        for (const {index, dictionary, dictionaryIndex, dictionaryPriority, character, frequency} of dictionaryEntry.frequencies) {
+        for (const {index, dictionary, dictionaryIndex, dictionaryPriority, character, frequency, displayValue, displayValueParsed} of dictionaryEntry.frequencies) {
             results.push({
                 index,
                 dictionary,
@@ -252,7 +252,7 @@ class AnkiNoteDataCreator {
                     priority: dictionaryPriority
                 },
                 character,
-                frequency
+                frequency: displayValueParsed ? displayValue : frequency
             });
         }
         return results;
@@ -374,7 +374,7 @@ class AnkiNoteDataCreator {
     _getTermFrequencies(dictionaryEntry) {
         const results = [];
         const {headwords} = dictionaryEntry;
-        for (const {headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency} of dictionaryEntry.frequencies) {
+        for (const {headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency, displayValue, displayValueParsed} of dictionaryEntry.frequencies) {
             const {term, reading} = headwords[headwordIndex];
             results.push({
                 index: results.length,
@@ -387,7 +387,7 @@ class AnkiNoteDataCreator {
                 expression: term,
                 reading,
                 hasReading,
-                frequency
+                frequency: displayValueParsed ? displayValue : frequency
             });
         }
         return results;
@@ -459,7 +459,7 @@ class AnkiNoteDataCreator {
     _getTermExpressionFrequencies(dictionaryEntry, i) {
         const results = [];
         const {headwords, frequencies} = dictionaryEntry;
-        for (const {headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency} of frequencies) {
+        for (const {headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency, displayValue, displayValueParsed} of frequencies) {
             if (headwordIndex !== i) { continue; }
             const {term, reading} = headwords[headwordIndex];
             results.push({
@@ -473,7 +473,7 @@ class AnkiNoteDataCreator {
                 expression: term,
                 reading,
                 hasReading,
-                frequency
+                frequency: displayValueParsed ? displayValue : frequency
             });
         }
         return results;
diff --git a/ext/js/display/display-generator.js b/ext/js/display/display-generator.js
index fe899e53..fc377df1 100644
--- a/ext/js/display/display-generator.js
+++ b/ext/js/display/display-generator.js
@@ -548,18 +548,15 @@ class DisplayGenerator {
 
         this._setTextContent(node.querySelector('.tag-label-content'), dictionary);
 
-        const frequency = values.join(', ');
-
         this._setTextContent(node.querySelector('.frequency-disambiguation-term'), term, 'ja');
         this._setTextContent(node.querySelector('.frequency-disambiguation-reading'), (reading !== null ? reading : ''), 'ja');
-        this._setTextContent(node.querySelector('.frequency-value'), frequency, 'ja');
+        this._populateFrequencyValueList(node.querySelector('.frequency-value-list'), values);
 
         node.dataset.term = term;
         node.dataset.reading = reading;
         node.dataset.hasReading = `${reading !== null}`;
         node.dataset.readingIsSame = `${reading === term}`;
         node.dataset.dictionary = dictionary;
-        node.dataset.frequency = `${frequency}`;
         node.dataset.details = dictionary;
 
         return node;
@@ -569,19 +566,50 @@ class DisplayGenerator {
         const {character, values} = details;
         const node = this._templates.instantiate('kanji-frequency-item');
 
-        const frequency = values.join(', ');
-
         this._setTextContent(node.querySelector('.tag-label-content'), dictionary);
-        this._setTextContent(node.querySelector('.frequency-value'), frequency, 'ja');
+        this._populateFrequencyValueList(node.querySelector('.frequency-value-list'), values);
 
         node.dataset.character = character;
         node.dataset.dictionary = dictionary;
-        node.dataset.frequency = `${frequency}`;
         node.dataset.details = dictionary;
 
         return node;
     }
 
+    _populateFrequencyValueList(node, values) {
+        let fullFrequency = '';
+        for (let i = 0, ii = values.length; i < ii; ++i) {
+            const {frequency, displayValue} = values[i];
+            const frequencyString = `${frequency}`;
+            const text = displayValue !== null ? displayValue : frequency;
+
+            if (i > 0) {
+                const node2 = document.createElement('span');
+                node2.className = 'frequency-value';
+                node2.dataset.frequency = `${frequency}`;
+                node2.textContent = ', ';
+                node.appendChild(node2);
+                fullFrequency += ', ';
+            }
+
+            const node2 = document.createElement('span');
+            node2.className = 'frequency-value';
+            node2.dataset.frequency = frequencyString;
+            if (displayValue !== null) {
+                node2.dataset.displayValue = `${displayValue}`;
+                if (displayValue !== frequencyString) {
+                    node2.title = frequencyString;
+                }
+            }
+            this._setTextContent(node2, text, 'ja');
+            node.appendChild(node2);
+
+            fullFrequency += text;
+        }
+
+        node.dataset.frequency = fullFrequency;
+    }
+
     _appendKanjiLinks(container, text) {
         const jp = this._japaneseUtil;
         let part = '';
diff --git a/ext/js/language/sandbox/dictionary-data-util.js b/ext/js/language/sandbox/dictionary-data-util.js
index 68b15c48..83d94b9b 100644
--- a/ext/js/language/sandbox/dictionary-data-util.js
+++ b/ext/js/language/sandbox/dictionary-data-util.js
@@ -48,7 +48,7 @@ class DictionaryDataUtil {
         const {headwords, frequencies} = dictionaryEntry;
 
         const map1 = new Map();
-        for (const {headwordIndex, dictionary, hasReading, frequency} of frequencies) {
+        for (const {headwordIndex, dictionary, hasReading, frequency, displayValue} of frequencies) {
             const {term, reading} = headwords[headwordIndex];
 
             let map2 = map1.get(dictionary);
@@ -61,18 +61,18 @@ class DictionaryDataUtil {
             const key = this._createMapKey([term, readingKey]);
             let frequencyData = map2.get(key);
             if (typeof frequencyData === 'undefined') {
-                frequencyData = {term, reading: readingKey, values: new Set()};
+                frequencyData = {term, reading: readingKey, values: new Map()};
                 map2.set(key, frequencyData);
             }
 
-            frequencyData.values.add(frequency);
+            frequencyData.values.set(this._createMapKey([frequency, displayValue]), {frequency, displayValue});
         }
         return this._createFrequencyGroupsFromMap(map1);
     }
 
     static groupKanjiFrequencies(frequencies) {
         const map1 = new Map();
-        for (const {dictionary, character, frequency} of frequencies) {
+        for (const {dictionary, character, frequency, displayValue} of frequencies) {
             let map2 = map1.get(dictionary);
             if (typeof map2 === 'undefined') {
                 map2 = new Map();
@@ -81,11 +81,11 @@ class DictionaryDataUtil {
 
             let frequencyData = map2.get(character);
             if (typeof frequencyData === 'undefined') {
-                frequencyData = {character, values: new Set()};
+                frequencyData = {character, values: new Map()};
                 map2.set(character, frequencyData);
             }
 
-            frequencyData.values.add(frequency);
+            frequencyData.values.set(this._createMapKey([frequency, displayValue]), {frequency, displayValue});
         }
         return this._createFrequencyGroupsFromMap(map1);
     }
@@ -222,7 +222,7 @@ class DictionaryDataUtil {
         for (const [dictionary, map2] of map.entries()) {
             const frequencies = [];
             for (const frequencyData of map2.values()) {
-                frequencyData.values = [...frequencyData.values];
+                frequencyData.values = [...frequencyData.values.values()];
                 frequencies.push(frequencyData);
             }
             results.push({dictionary, frequencies});
diff --git a/ext/js/language/translator.js b/ext/js/language/translator.js
index 056ff3a7..28e1cfcc 100644
--- a/ext/js/language/translator.js
+++ b/ext/js/language/translator.js
@@ -36,6 +36,7 @@ class Translator {
         this._deinflector = null;
         this._tagCache = new Map();
         this._stringComparer = new Intl.Collator('en-US'); // Invariant locale
+        this._numberRegex = /[+-]?(\d+(\.\d*)?|\.\d+)([eE][+-]?\d+)?/;
     }
 
     /**
@@ -853,12 +854,15 @@ class Translator {
                     case 'freq':
                         {
                             let frequency = data;
-                            const hasReading = (data !== null && typeof data === 'object');
+                            const hasReading = (data !== null && typeof data === 'object' && typeof data.reading === 'string');
                             if (hasReading) {
                                 if (data.reading !== reading) { continue; }
                                 frequency = data.frequency;
                             }
                             for (const {frequencies, headwordIndex} of targets) {
+                                let displayValue;
+                                let displayValueParsed;
+                                ({frequency, displayValue, displayValueParsed} = this._getFrequencyInfo(frequency));
                                 frequencies.push(this._createTermFrequency(
                                     frequencies.length,
                                     headwordIndex,
@@ -866,7 +870,9 @@ class Translator {
                                     dictionaryIndex,
                                     dictionaryPriority,
                                     hasReading,
-                                    this._convertFrequency(frequency)
+                                    frequency,
+                                    displayValue,
+                                    displayValueParsed
                                 ));
                             }
                         }
@@ -914,13 +920,16 @@ class Translator {
                 case 'freq':
                     {
                         const {frequencies} = dictionaryEntries[index];
+                        const {frequency, displayValue, displayValueParsed} = this._getFrequencyInfo(data);
                         frequencies.push(this._createKanjiFrequency(
                             frequencies.length,
                             dictionary,
                             dictionaryIndex,
                             dictionaryPriority,
                             character,
-                            this._convertFrequency(data)
+                            frequency,
+                            displayValue,
+                            displayValueParsed
                         ));
                     }
                     break;
@@ -971,16 +980,36 @@ class Translator {
         });
     }
 
-    _convertFrequency(value) {
-        switch (typeof value) {
-            case 'number':
-                return value;
-            case 'string':
-                value = Number.parseFloat(value);
-                return Number.isFinite(value) ? value : 0;
-            default:
-                return 0;
+    _convertStringToNumber(value) {
+        const match = this._numberRegex.exec(value);
+        if (match === null) { return 0; }
+        value = Number.parseFloat(match[0]);
+        return Number.isFinite(value) ? value : 0;
+    }
+
+    _getFrequencyInfo(frequency) {
+        let displayValue = null;
+        let displayValueParsed = false;
+        if (typeof frequency === 'object' && frequency !== null) {
+            ({value: frequency, displayValue} = frequency);
+            if (typeof frequency !== 'number') { frequency = 0; }
+            if (typeof displayValue !== 'string') { displayValue = null; }
+        } else {
+            switch (typeof frequency) {
+                case 'number':
+                    // No change
+                    break;
+                case 'string':
+                    displayValue = frequency;
+                    displayValueParsed = true;
+                    frequency = this._convertStringToNumber(frequency);
+                    break;
+                default:
+                    frequency = 0;
+                    break;
+            }
         }
+        return {frequency, displayValue, displayValueParsed};
     }
 
     // Helpers
@@ -1048,8 +1077,8 @@ class Translator {
         };
     }
 
-    _createKanjiFrequency(index, dictionary, dictionaryIndex, dictionaryPriority, character, frequency) {
-        return {index, dictionary, dictionaryIndex, dictionaryPriority, character, frequency};
+    _createKanjiFrequency(index, dictionary, dictionaryIndex, dictionaryPriority, character, frequency, displayValue, displayValueParsed) {
+        return {index, dictionary, dictionaryIndex, dictionaryPriority, character, frequency, displayValue, displayValueParsed};
     }
 
     _createKanjiDictionaryEntry(character, dictionary, onyomi, kunyomi, tags, stats, definitions) {
@@ -1114,8 +1143,8 @@ class Translator {
         return {index, headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, pitches};
     }
 
-    _createTermFrequency(index, headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency) {
-        return {index, headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency};
+    _createTermFrequency(index, headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency, displayValue, displayValueParsed) {
+        return {index, headwordIndex, dictionary, dictionaryIndex, dictionaryPriority, hasReading, frequency, displayValue, displayValueParsed};
     }
 
     _createTermDictionaryEntry(isPrimary, inflections, score, dictionaryIndex, dictionaryPriority, sourceTermExactMatchCount, maxTransformedTextLength, headwords, definitions) {