@mailwoman/classifiers 2.0.0

package/out/AdjacencyClassifier.d.ts

@@ -0,0 +1,16 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { SectionClassifier, Span } from "@mailwoman/core";
+/**
+ * An adjacency classifier is used to find three adjacent words in a section.
+ *
+ * The first word must be a house number, the second word must be a street, and the third word must
+ * be a street suffix.
+ */
+export declare class AdjacencyClassifier extends SectionClassifier {
+    explore(section: Span): void;
+}
+//# sourceMappingURL=AdjacencyClassifier.d.ts.map

package/out/AdjacencyClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"AdjacencyClassifier.d.ts","sourceRoot":"","sources":["../AdjacencyClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,iBAAiB,EAAE,IAAI,EAAE,MAAM,iBAAiB,CAAA;AAEzD;;;;;GAKG;AACH,qBAAa,mBAAoB,SAAQ,iBAAiB;IACzD,OAAO,CAAC,OAAO,EAAE,IAAI,GAAG,IAAI;CA6C5B"}

package/out/AdjacencyClassifier.js

@@ -0,0 +1,50 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { SectionClassifier, Span } from "@mailwoman/core";
+/**
+ * An adjacency classifier is used to find three adjacent words in a section.
+ *
+ * The first word must be a house number, the second word must be a street, and the third word must
+ * be a street suffix.
+ */
+export class AdjacencyClassifier extends SectionClassifier {
+    explore(section) {
+        Array.from(section.children).forEach((primaryChild, i, children) => {
+            // skip last two elements
+            if (i >= section.children.size - 2)
+                return;
+            const secondaryChild = children[i + 1];
+            const tertiaryChild = children[i + 2];
+            const primaryChildHasHouseNumber = primaryChild.is("house_number");
+            const secondaryChildHasStreet = secondaryChild.is("street");
+            const secondaryChildPhraseHasStreet = SectionClassifier.findPhrasesContaining(section, secondaryChild).some((p) => {
+                return p.is("street");
+            });
+            const tertiaryChildHasStreetSuffix = tertiaryChild.is("street_suffix");
+            if (primaryChildHasHouseNumber &&
+                (secondaryChildHasStreet || secondaryChildPhraseHasStreet) &&
+                tertiaryChildHasStreetSuffix) {
+                // Every child must be part of the set above and must not omit any children.
+                const matches = Iterator.from(section.phrases).filter((phrase) => {
+                    if (phrase.children.size !== 3)
+                        return false;
+                    const [primaryPhrase, secondPhrase, tertiaryPhrase] = phrase.children;
+                    return (
+                    // ---
+                    primaryPhrase === primaryChild &&
+                        // ---
+                        secondPhrase === secondaryChild &&
+                        // ---
+                        tertiaryPhrase === tertiaryChild);
+                });
+                for (const match of matches) {
+                    match.classifications.add("adjacent");
+                }
+            }
+        });
+    }
+}
+//# sourceMappingURL=AdjacencyClassifier.js.map

package/out/AdjacencyClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"AdjacencyClassifier.js","sourceRoot":"","sources":["../AdjacencyClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,iBAAiB,EAAE,IAAI,EAAE,MAAM,iBAAiB,CAAA;AAEzD;;;;;GAKG;AACH,MAAM,OAAO,mBAAoB,SAAQ,iBAAiB;IACzD,OAAO,CAAC,OAAa;QACpB,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,OAAO,CAAC,CAAC,YAAY,EAAE,CAAC,EAAE,QAAyB,EAAE,EAAE;YACnF,yBAAyB;YACzB,IAAI,CAAC,IAAI,OAAO,CAAC,QAAQ,CAAC,IAAI,GAAG,CAAC;gBAAE,OAAM;YAE1C,MAAM,cAAc,GAAG,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAE,CAAA;YACvC,MAAM,aAAa,GAAG,QAAQ,CAAC,CAAC,GAAG,CAAC,CAAE,CAAA;YAEtC,MAAM,0BAA0B,GAAG,YAAY,CAAC,EAAE,CAAC,cAAc,CAAC,CAAA;YAClE,MAAM,uBAAuB,GAAG,cAAc,CAAC,EAAE,CAAC,QAAQ,CAAC,CAAA;YAC3D,MAAM,6BAA6B,GAAG,iBAAiB,CAAC,qBAAqB,CAAC,OAAO,EAAE,cAAc,CAAC,CAAC,IAAI,CAC1G,CAAC,CAAC,EAAE,EAAE;gBACL,OAAO,CAAC,CAAC,EAAE,CAAC,QAAQ,CAAC,CAAA;YACtB,CAAC,CACD,CAAA;YAED,MAAM,4BAA4B,GAAG,aAAa,CAAC,EAAE,CAAC,eAAe,CAAC,CAAA;YAEtE,IACC,0BAA0B;gBAC1B,CAAC,uBAAuB,IAAI,6BAA6B,CAAC;gBAC1D,4BAA4B,EAC3B,CAAC;gBACF,4EAA4E;gBAC5E,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,MAAM,EAAE,EAAE;oBAChE,IAAI,MAAM,CAAC,QAAQ,CAAC,IAAI,KAAK,CAAC;wBAAE,OAAO,KAAK,CAAA;oBAE5C,MAAM,CAAC,aAAa,EAAE,YAAY,EAAE,cAAc,CAAC,GAAG,MAAM,CAAC,QAAQ,CAAA;oBAErE,OAAO;oBACN,MAAM;oBACN,aAAa,KAAK,YAAY;wBAC9B,MAAM;wBACN,YAAY,KAAK,cAAc;wBAC/B,MAAM;wBACN,cAAc,KAAK,aAAa,CAChC,CAAA;gBACF,CAAC,CAAC,CAAA;gBAEF,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;oBAC7B,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;gBACtC,CAAC;YACF,CAAC;QACF,CAAC,CAAC,CAAA;IACH,CAAC;CACD"}

package/out/AlphaNumericClassifier.d.ts

@@ -0,0 +1,10 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, WordClassifier } from "@mailwoman/core";
+export declare class AlphaNumericClassifier extends WordClassifier {
+    explore(span: Span): void;
+}
+//# sourceMappingURL=AlphaNumericClassifier.d.ts.map

package/out/AlphaNumericClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"AlphaNumericClassifier.d.ts","sourceRoot":"","sources":["../AlphaNumericClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAEtD,qBAAa,sBAAuB,SAAQ,cAAc;IAClD,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI;CAWhC"}

package/out/AlphaNumericClassifier.js

@@ -0,0 +1,23 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, WordClassifier } from "@mailwoman/core";
+export class AlphaNumericClassifier extends WordClassifier {
+    explore(span) {
+        if (span.flags.has("numeric")) {
+            span.classifications.add("numeric");
+        }
+        else if (span.flags.has("numeral")) {
+            span.classifications.add("alphanumeric");
+        }
+        else if (span.flags.has("punctuation")) {
+            span.classifications.add("punctuation");
+        }
+        else {
+            span.classifications.add("alpha");
+        }
+    }
+}
+//# sourceMappingURL=AlphaNumericClassifier.js.map

package/out/AlphaNumericClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"AlphaNumericClassifier.js","sourceRoot":"","sources":["../AlphaNumericClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAEtD,MAAM,OAAO,sBAAuB,SAAQ,cAAc;IAClD,OAAO,CAAC,IAAU;QACxB,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;YAC/B,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QACpC,CAAC;aAAM,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC;YACtC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,cAAc,CAAC,CAAA;QACzC,CAAC;aAAM,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,aAAa,CAAC,EAAE,CAAC;YAC1C,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,aAAa,CAAC,CAAA;QACxC,CAAC;aAAM,CAAC;YACP,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;QAClC,CAAC;IACF,CAAC;CACD"}

package/out/CentralEuropeanStreetNameClassifier.d.ts

@@ -0,0 +1,16 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { SectionClassifier, Span } from "@mailwoman/core";
+/**
+ * Classifies street names with no suffix or prefix when accompanied by a house number in the same
+ * section.
+ *
+ * @see {@link https://github.com/pelias/parser/issues/83 | Issue #83}
+ */
+export declare class CentralEuropeanStreetNameClassifier extends SectionClassifier {
+    explore(section: Span): void;
+}
+//# sourceMappingURL=CentralEuropeanStreetNameClassifier.d.ts.map

package/out/CentralEuropeanStreetNameClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CentralEuropeanStreetNameClassifier.d.ts","sourceRoot":"","sources":["../CentralEuropeanStreetNameClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,iBAAiB,EAAE,IAAI,EAAE,MAAM,iBAAiB,CAAA;AAEzD;;;;;GAKG;AACH,qBAAa,mCAAoC,SAAQ,iBAAiB;IAClE,OAAO,CAAC,OAAO,EAAE,IAAI,GAAG,IAAI;CAsBnC"}

package/out/CentralEuropeanStreetNameClassifier.js

@@ -0,0 +1,36 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { SectionClassifier, Span } from "@mailwoman/core";
+/**
+ * Classifies street names with no suffix or prefix when accompanied by a house number in the same
+ * section.
+ *
+ * @see {@link https://github.com/pelias/parser/issues/83 | Issue #83}
+ */
+export class CentralEuropeanStreetNameClassifier extends SectionClassifier {
+    explore(section) {
+        const { first, last } = section.children;
+        // Does the section have at least two children?
+        if (!first || !last)
+            return;
+        // Is the first child a toponym, like a region or city?
+        if (first.is("toponym"))
+            return;
+        // Section doesn't end with a housenumber?
+        if (!last.is("house_number"))
+            return;
+        // Other elements cannot contain any public classifications.
+        if (first.classifications.hasVisibleClassification())
+            return;
+        // Assume the first token is a street name...
+        first.classifications.add({
+            classification: "street",
+            confidence: 0.5,
+            flags: new Set(["central_european_street_name"]),
+        });
+    }
+}
+//# sourceMappingURL=CentralEuropeanStreetNameClassifier.js.map

package/out/CentralEuropeanStreetNameClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CentralEuropeanStreetNameClassifier.js","sourceRoot":"","sources":["../CentralEuropeanStreetNameClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,iBAAiB,EAAE,IAAI,EAAE,MAAM,iBAAiB,CAAA;AAEzD;;;;;GAKG;AACH,MAAM,OAAO,mCAAoC,SAAQ,iBAAiB;IAClE,OAAO,CAAC,OAAa;QAC3B,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,OAAO,CAAC,QAAQ,CAAA;QAExC,+CAA+C;QAC/C,IAAI,CAAC,KAAK,IAAI,CAAC,IAAI;YAAE,OAAM;QAE3B,uDAAuD;QACvD,IAAI,KAAK,CAAC,EAAE,CAAC,SAAS,CAAC;YAAE,OAAM;QAE/B,0CAA0C;QAC1C,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,cAAc,CAAC;YAAE,OAAM;QAEpC,4DAA4D;QAC5D,IAAI,KAAK,CAAC,eAAe,CAAC,wBAAwB,EAAE;YAAE,OAAM;QAE5D,6CAA6C;QAC7C,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC;YACzB,cAAc,EAAE,QAAQ;YACxB,UAAU,EAAE,GAAG;YACf,KAAK,EAAE,IAAI,GAAG,CAAC,CAAC,8BAA8B,CAAC,CAAC;SAChD,CAAC,CAAA;IACH,CAAC;CACD"}

package/out/ChainClassifier.d.ts

@@ -0,0 +1,14 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { PhraseClassifier, Span } from "@mailwoman/core";
+/**
+ * Chain-restaurants and other franchises.
+ */
+export declare class ChainClassifier extends PhraseClassifier {
+    ready(): Promise<this>;
+    explore(span: Span): void;
+}
+//# sourceMappingURL=ChainClassifier.d.ts.map

package/out/ChainClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"ChainClassifier.d.ts","sourceRoot":"","sources":["../ChainClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,gBAAgB,EAAE,IAAI,EAAsB,MAAM,iBAAiB,CAAA;AAE5E;;GAEG;AACH,qBAAa,eAAgB,SAAQ,gBAAgB;IACvC,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAM5B,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI;CAUhC"}

package/out/ChainClassifier.js

@@ -0,0 +1,25 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { PhraseClassifier, Span, prepareLocaleIndex } from "@mailwoman/core";
+/**
+ * Chain-restaurants and other franchises.
+ */
+export class ChainClassifier extends PhraseClassifier {
+    async ready() {
+        this.index = await prepareLocaleIndex(this.languages, "chains.txt");
+        return this;
+    }
+    explore(span) {
+        const languages = this.index.get(span.normalized);
+        if (languages) {
+            span.classifications.add({
+                classification: "chain",
+                languages,
+            });
+        }
+    }
+}
+//# sourceMappingURL=ChainClassifier.js.map

package/out/ChainClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ChainClassifier.js","sourceRoot":"","sources":["../ChainClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,gBAAgB,EAAE,IAAI,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAA;AAE5E;;GAEG;AACH,MAAM,OAAO,eAAgB,SAAQ,gBAAgB;IAC7C,KAAK,CAAC,KAAK;QACjB,IAAI,CAAC,KAAK,GAAG,MAAM,kBAAkB,CAAC,IAAI,CAAC,SAAS,EAAE,YAAY,CAAC,CAAA;QAEnE,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,OAAO,CAAC,IAAU;QACxB,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;QAEjD,IAAI,SAAS,EAAE,CAAC;YACf,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC;gBACxB,cAAc,EAAE,OAAO;gBACvB,SAAS;aACT,CAAC,CAAA;QACH,CAAC;IACF,CAAC;CACD"}

package/out/CompoundLevelClassifier.d.ts

@@ -0,0 +1,16 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { type Classifier, type ClassifierOptions, type LibPostalLanguageCode, LocaleIndex, Span, TokenContext } from "@mailwoman/core";
+export declare class CompoundLevelClassifier implements Classifier {
+    index: LocaleIndex<LibPostalLanguageCode>;
+    protected languages?: LibPostalLanguageCode[];
+    constructor({ languages }?: ClassifierOptions);
+    ready(): Promise<this>;
+    classifyTokens(context: TokenContext): void;
+    classify(input: Span | string): Span;
+    explore(span: Span, section: Span): void;
+}
+//# sourceMappingURL=CompoundLevelClassifier.d.ts.map

package/out/CompoundLevelClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompoundLevelClassifier.d.ts","sourceRoot":"","sources":["../CompoundLevelClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAEN,KAAK,UAAU,EACf,KAAK,iBAAiB,EACtB,KAAK,qBAAqB,EAC1B,WAAW,EAEX,IAAI,EACJ,YAAY,EACZ,MAAM,iBAAiB,CAAA;AAExB,qBAAa,uBAAwB,YAAW,UAAU;IAClD,KAAK,EAAG,WAAW,CAAC,qBAAqB,CAAC,CAAA;IACjD,SAAS,CAAC,SAAS,CAAC,EAAE,qBAAqB,EAAE,CAAA;gBAEjC,EAAE,SAAS,EAAE,GAAE,iBAAsB;IAIpC,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAO5B,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,IAAI;IAU3C,QAAQ,CAAC,KAAK,EAAE,IAAI,GAAG,MAAM,GAAG,IAAI;IAQpC,OAAO,CAAC,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,GAAG,IAAI;CAkC/C"}

package/out/CompoundLevelClassifier.js

@@ -0,0 +1,58 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Alpha2LanguageCode, LocaleIndex, prepareLocaleIndex, Span, TokenContext, } from "@mailwoman/core";
+export class CompoundLevelClassifier {
+    index;
+    languages;
+    constructor({ languages } = {}) {
+        this.languages = languages ? Array.from(languages) : undefined;
+    }
+    async ready() {
+        const languages = this.languages ?? [Alpha2LanguageCode.English];
+        this.index = await prepareLocaleIndex(languages, "level_types_numbered.txt");
+        return this;
+    }
+    classifyTokens(context) {
+        for (let sectionIndex = 0; sectionIndex < context.sections.length; sectionIndex++) {
+            const { children } = context.sections[sectionIndex];
+            for (const child of children) {
+                this.explore(child, context.sections[sectionIndex]);
+            }
+        }
+    }
+    classify(input) {
+        const span = Span.from(input);
+        this.explore(span, span);
+        return span;
+    }
+    explore(span, section) {
+        if (!span.flags.has("alphanumeric"))
+            return;
+        // We a searching spans like `F12` which means `Floor 12`
+        for (const token of this.index.keys()) {
+            if (span.body.length < token.length)
+                continue;
+            if (span.normalized.substring(0, token.length) === token &&
+                /^\d+$/.test(span.normalized.substring(token.length))) {
+                const levelTypeBody = span.body.substring(0, token.length);
+                const levelBody = span.body.substring(token.length);
+                const levelType = Span.from(levelTypeBody, { start: span.start });
+                const level = Span.from(levelBody, { start: span.start + levelTypeBody.length });
+                // We are creating two spans `{level_type} {level}`
+                levelType.classifications.add("level_designator");
+                levelType.nextSiblings.add(level);
+                level.classifications.add("level");
+                level.previousSiblings.add(levelType);
+                span.previousSiblings.forEach((previousSibling) => levelType.previousSiblings.add(previousSibling));
+                span.nextSiblings.forEach((nextSibling) => level.nextSiblings.add(nextSibling));
+                section.children.add(levelType);
+                section.children.add(level);
+                return;
+            }
+        }
+    }
+}
+//# sourceMappingURL=CompoundLevelClassifier.js.map

package/out/CompoundLevelClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompoundLevelClassifier.js","sourceRoot":"","sources":["../CompoundLevelClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EACN,kBAAkB,EAIlB,WAAW,EACX,kBAAkB,EAClB,IAAI,EACJ,YAAY,GACZ,MAAM,iBAAiB,CAAA;AAExB,MAAM,OAAO,uBAAuB;IAC5B,KAAK,CAAqC;IACvC,SAAS,CAA0B;IAE7C,YAAY,EAAE,SAAS,KAAwB,EAAE;QAChD,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IAC/D,CAAC;IAEM,KAAK,CAAC,KAAK;QACjB,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,IAAI,CAAC,kBAAkB,CAAC,OAAO,CAAC,CAAA;QAChE,IAAI,CAAC,KAAK,GAAG,MAAM,kBAAkB,CAAC,SAAS,EAAE,0BAA0B,CAAC,CAAA;QAE5E,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,cAAc,CAAC,OAAqB;QAC1C,KAAK,IAAI,YAAY,GAAG,CAAC,EAAE,YAAY,GAAG,OAAO,CAAC,QAAQ,CAAC,MAAM,EAAE,YAAY,EAAE,EAAE,CAAC;YACnF,MAAM,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC,QAAQ,CAAC,YAAY,CAAE,CAAA;YAEpD,KAAK,MAAM,KAAK,IAAI,QAAQ,EAAE,CAAC;gBAC9B,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,OAAO,CAAC,QAAQ,CAAC,YAAY,CAAE,CAAC,CAAA;YACrD,CAAC;QACF,CAAC;IACF,CAAC;IAEM,QAAQ,CAAC,KAAoB;QACnC,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAE7B,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;QAExB,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,OAAO,CAAC,IAAU,EAAE,OAAa;QACvC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,cAAc,CAAC;YAAE,OAAM;QAE3C,yDAAyD;QACzD,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;YACvC,IAAI,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM;gBAAE,SAAQ;YAE7C,IACC,IAAI,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,KAAK,KAAK;gBACpD,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,EACpD,CAAC;gBACF,MAAM,aAAa,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAA;gBAC1D,MAAM,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;gBAEnD,MAAM,SAAS,GAAG,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,CAAC,CAAA;gBACjE,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,GAAG,aAAa,CAAC,MAAM,EAAE,CAAC,CAAA;gBAEhF,mDAAmD;gBACnD,SAAS,CAAC,eAAe,CAAC,GAAG,CAAC,kBAAkB,CAAC,CAAA;gBACjD,SAAS,CAAC,YAAY,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;gBAEjC,KAAK,CAAC,eAAe,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;gBAClC,KAAK,CAAC,gBAAgB,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;gBAErC,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,CAAC,eAAe,EAAE,EAAE,CAAC,SAAS,CAAC,gBAAgB,CAAC,GAAG,CAAC,eAAe,CAAC,CAAC,CAAA;gBACnG,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,WAAW,EAAE,EAAE,CAAC,KAAK,CAAC,YAAY,CAAC,GAAG,CAAC,WAAW,CAAC,CAAC,CAAA;gBAE/E,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;gBAC/B,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;gBAE3B,OAAM;YACP,CAAC;QACF,CAAC;IACF,CAAC;CACD"}

package/out/CompoundStreetClassifier.d.ts

@@ -0,0 +1,12 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { type LibPostalLanguageCode, LocaleIndex, Span, WordClassifier } from "@mailwoman/core";
+export declare class CompoundStreetClassifier extends WordClassifier {
+    suffixes: LocaleIndex<LibPostalLanguageCode>;
+    ready(): Promise<this>;
+    explore(span: Span): void;
+}
+//# sourceMappingURL=CompoundStreetClassifier.d.ts.map

package/out/CompoundStreetClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompoundStreetClassifier.d.ts","sourceRoot":"","sources":["../CompoundStreetClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAEN,KAAK,qBAAqB,EAC1B,WAAW,EAEX,IAAI,EACJ,cAAc,EACd,MAAM,iBAAiB,CAAA;AAExB,qBAAa,wBAAyB,SAAQ,cAAc;IACpD,QAAQ,EAAG,WAAW,CAAC,qBAAqB,CAAC,CAAA;IAE9C,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAgCrB,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI;CAqBhC"}

package/out/CompoundStreetClassifier.js

@@ -0,0 +1,50 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Alpha2LanguageCode, LocaleIndex, prepareLocaleIndex, Span, WordClassifier, } from "@mailwoman/core";
+export class CompoundStreetClassifier extends WordClassifier {
+    suffixes;
+    async ready() {
+        const [separable, inseparable] = await Promise.all([
+            prepareLocaleIndex([
+                Alpha2LanguageCode.German,
+                Alpha2LanguageCode.Dutch,
+                Alpha2LanguageCode.Swedish,
+                Alpha2LanguageCode.NorwegianBokmål,
+            ], "concatenated_suffixes_separable.txt", {
+                // Removes suffixes such as 'r.' which can be ambiguous
+                minLength: 3,
+            }),
+            prepareLocaleIndex([Alpha2LanguageCode.German, Alpha2LanguageCode.Dutch, Alpha2LanguageCode.NorwegianBokmål], "concatenated_suffixes_inseparable.txt", {
+                // Removes suffixes such as 'r.' which can be ambiguous
+                minLength: 3,
+            }),
+        ]);
+        this.suffixes = new LocaleIndex([...separable, ...inseparable], {
+            displayName: "libpostal",
+        });
+        return this;
+    }
+    explore(span) {
+        if (span.flags.has("numeral"))
+            return;
+        // else use a slower suffix check which is O(n)
+        // this allows us to match Germanic compound words such as:
+        // 'Grolmanstraße' which end with the dictionary term '-straße'
+        for (const [token, languages] of this.suffixes) {
+            const offset = span.body.length - token.length;
+            if (offset < 1)
+                continue;
+            if (span.normalized.substring(offset) === token) {
+                span.classifications.add({
+                    classification: "street",
+                    languages,
+                });
+                return;
+            }
+        }
+    }
+}
+//# sourceMappingURL=CompoundStreetClassifier.js.map

package/out/CompoundStreetClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompoundStreetClassifier.js","sourceRoot":"","sources":["../CompoundStreetClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EACN,kBAAkB,EAElB,WAAW,EACX,kBAAkB,EAClB,IAAI,EACJ,cAAc,GACd,MAAM,iBAAiB,CAAA;AAExB,MAAM,OAAO,wBAAyB,SAAQ,cAAc;IACpD,QAAQ,CAAqC;IAEpD,KAAK,CAAC,KAAK;QACV,MAAM,CAAC,SAAS,EAAE,WAAW,CAAC,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC;YAClD,kBAAkB,CACjB;gBACC,kBAAkB,CAAC,MAAM;gBACzB,kBAAkB,CAAC,KAAK;gBACxB,kBAAkB,CAAC,OAAO;gBAC1B,kBAAkB,CAAC,eAAe;aAClC,EACD,qCAAqC,EACrC;gBACC,uDAAuD;gBACvD,SAAS,EAAE,CAAC;aACZ,CACD;YACD,kBAAkB,CACjB,CAAC,kBAAkB,CAAC,MAAM,EAAE,kBAAkB,CAAC,KAAK,EAAE,kBAAkB,CAAC,eAAe,CAAC,EACzF,uCAAuC,EACvC;gBACC,uDAAuD;gBACvD,SAAS,EAAE,CAAC;aACZ,CACD;SACD,CAAC,CAAA;QAEF,IAAI,CAAC,QAAQ,GAAG,IAAI,WAAW,CAAwB,CAAC,GAAG,SAAS,EAAE,GAAG,WAAW,CAAC,EAAE;YACtF,WAAW,EAAE,WAAW;SACxB,CAAC,CAAA;QAEF,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,OAAO,CAAC,IAAU;QACxB,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC;YAAE,OAAM;QAErC,+CAA+C;QAC/C,2DAA2D;QAC3D,+DAA+D;QAC/D,KAAK,MAAM,CAAC,KAAK,EAAE,SAAS,CAAC,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;YAChD,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,CAAA;YAE9C,IAAI,MAAM,GAAG,CAAC;gBAAE,SAAQ;YAExB,IAAI,IAAI,CAAC,UAAU,CAAC,SAAS,CAAC,MAAM,CAAC,KAAK,KAAK,EAAE,CAAC;gBACjD,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC;oBACxB,cAAc,EAAE,QAAQ;oBACxB,SAAS;iBACT,CAAC,CAAA;gBAEF,OAAM;YACP,CAAC;QACF,CAAC;IACF,CAAC;CACD"}

package/out/CompoundUnitDesignatorClassifier.d.ts

@@ -0,0 +1,16 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { type Classifier, type ClassifierOptions, type LibPostalLanguageCode, LocaleIndex, Span, TokenContext } from "@mailwoman/core";
+export declare class CompoundUnitDesignatorClassifier implements Classifier {
+    index: LocaleIndex<LibPostalLanguageCode>;
+    protected languages?: LibPostalLanguageCode[];
+    constructor({ languages }?: ClassifierOptions);
+    ready(): Promise<this>;
+    classifyTokens(context: TokenContext): void;
+    classify(input: Span | string): Span;
+    explore(span: Span, section: Span): void;
+}
+//# sourceMappingURL=CompoundUnitDesignatorClassifier.d.ts.map

package/out/CompoundUnitDesignatorClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompoundUnitDesignatorClassifier.d.ts","sourceRoot":"","sources":["../CompoundUnitDesignatorClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAEN,KAAK,UAAU,EACf,KAAK,iBAAiB,EACtB,KAAK,qBAAqB,EAC1B,WAAW,EAEX,IAAI,EACJ,YAAY,EACZ,MAAM,iBAAiB,CAAA;AAExB,qBAAa,gCAAiC,YAAW,UAAU;IAC3D,KAAK,EAAG,WAAW,CAAC,qBAAqB,CAAC,CAAA;IACjD,SAAS,CAAC,SAAS,CAAC,EAAE,qBAAqB,EAAE,CAAA;gBAEjC,EAAE,SAAS,EAAE,GAAE,iBAAsB;IAIpC,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAO5B,cAAc,CAAC,OAAO,EAAE,YAAY,GAAG,IAAI;IAQ3C,QAAQ,CAAC,KAAK,EAAE,IAAI,GAAG,MAAM,GAAG,IAAI;IAQpC,OAAO,CAAC,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,GAAG,IAAI;CAkC/C"}

package/out/CompoundUnitDesignatorClassifier.js

@@ -0,0 +1,57 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Alpha2LanguageCode, LocaleIndex, prepareLocaleIndex, Span, TokenContext, } from "@mailwoman/core";
+export class CompoundUnitDesignatorClassifier {
+    index;
+    languages;
+    constructor({ languages } = {}) {
+        this.languages = languages ? Array.from(languages) : undefined;
+    }
+    async ready() {
+        const languages = this.languages ?? [Alpha2LanguageCode.English];
+        this.index = await prepareLocaleIndex(languages, "unit_types_numbered.txt");
+        return this;
+    }
+    classifyTokens(context) {
+        for (const section of context.sections) {
+            for (const child of section.children) {
+                this.explore(child, section);
+            }
+        }
+    }
+    classify(input) {
+        const span = Span.from(input);
+        this.explore(span, span);
+        return span;
+    }
+    explore(span, section) {
+        if (!span.flags.has("alphanumeric"))
+            return;
+        // We a searching spans like `U12` which means `Unit 12`
+        for (const token of this.index.keys()) {
+            if (span.body.length < token.length)
+                continue;
+            if (span.normalized.substring(0, token.length) === token &&
+                /^\d+$/.test(span.normalized.substring(token.length))) {
+                const unitTypeBody = span.body.substring(0, token.length);
+                const unitBody = span.body.substring(token.length);
+                const unitType = Span.from(unitTypeBody, { start: span.start });
+                const unit = Span.from(unitBody, { start: span.start + unitTypeBody.length });
+                // We are creating two spans `{unit_type} {unit}`
+                unitType.classifications.add("unit_designator");
+                unitType.nextSiblings.add(unit);
+                unit.classifications.add("unit");
+                unit.previousSiblings.add(unitType);
+                span.previousSiblings.forEach((prev) => unitType.previousSiblings.add(prev));
+                span.nextSiblings.forEach((next) => unit.nextSiblings.add(next));
+                section.children.add(unitType);
+                section.children.add(unit);
+                return;
+            }
+        }
+    }
+}
+//# sourceMappingURL=CompoundUnitDesignatorClassifier.js.map

package/out/CompoundUnitDesignatorClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"CompoundUnitDesignatorClassifier.js","sourceRoot":"","sources":["../CompoundUnitDesignatorClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EACN,kBAAkB,EAIlB,WAAW,EACX,kBAAkB,EAClB,IAAI,EACJ,YAAY,GACZ,MAAM,iBAAiB,CAAA;AAExB,MAAM,OAAO,gCAAgC;IACrC,KAAK,CAAqC;IACvC,SAAS,CAA0B;IAE7C,YAAY,EAAE,SAAS,KAAwB,EAAE;QAChD,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IAC/D,CAAC;IAEM,KAAK,CAAC,KAAK;QACjB,MAAM,SAAS,GAAG,IAAI,CAAC,SAAS,IAAI,CAAC,kBAAkB,CAAC,OAAO,CAAC,CAAA;QAChE,IAAI,CAAC,KAAK,GAAG,MAAM,kBAAkB,CAAC,SAAS,EAAE,yBAAyB,CAAC,CAAA;QAE3E,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,cAAc,CAAC,OAAqB;QAC1C,KAAK,MAAM,OAAO,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;YACxC,KAAK,MAAM,KAAK,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;gBACtC,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,OAAO,CAAC,CAAA;YAC7B,CAAC;QACF,CAAC;IACF,CAAC;IAEM,QAAQ,CAAC,KAAoB;QACnC,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QAE7B,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;QAExB,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,OAAO,CAAC,IAAU,EAAE,OAAa;QACvC,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,cAAc,CAAC;YAAE,OAAM;QAE3C,wDAAwD;QACxD,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;YACvC,IAAI,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM;gBAAE,SAAQ;YAE7C,IACC,IAAI,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,KAAK,KAAK;gBACpD,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,EACpD,CAAC;gBACF,MAAM,YAAY,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,CAAC,CAAA;gBACzD,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;gBAElD,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,YAAY,EAAE,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,CAAC,CAAA;gBAC/D,MAAM,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,GAAG,YAAY,CAAC,MAAM,EAAE,CAAC,CAAA;gBAE7E,iDAAiD;gBACjD,QAAQ,CAAC,eAAe,CAAC,GAAG,CAAC,iBAAiB,CAAC,CAAA;gBAC/C,QAAQ,CAAC,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;gBAE/B,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;gBAChC,IAAI,CAAC,gBAAgB,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;gBAEnC,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAA;gBAC5E,IAAI,CAAC,YAAY,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAA;gBAEhE,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;gBAC9B,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;gBAE1B,OAAM;YACP,CAAC;QACF,CAAC;IACF,CAAC;CACD"}

package/out/DirectionalClassifier.d.ts

@@ -0,0 +1,11 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, WordClassifier } from "@mailwoman/core";
+export declare class DirectionalClassifier extends WordClassifier {
+    ready(): Promise<this>;
+    explore(span: Span): void;
+}
+//# sourceMappingURL=DirectionalClassifier.d.ts.map

package/out/DirectionalClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"DirectionalClassifier.d.ts","sourceRoot":"","sources":["../DirectionalClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAsB,IAAI,EAAE,cAAc,EAAsB,MAAM,iBAAiB,CAAA;AAE9F,qBAAa,qBAAsB,SAAQ,cAAc;IAC3C,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAgB5B,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI;CAWhC"}

package/out/DirectionalClassifier.js

@@ -0,0 +1,31 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Alpha2LanguageCode, Span, WordClassifier, prepareLocaleIndex } from "@mailwoman/core";
+export class DirectionalClassifier extends WordClassifier {
+    async ready() {
+        this.index = await prepareLocaleIndex(this.languages || [
+            Alpha2LanguageCode.English,
+            Alpha2LanguageCode.Spanish,
+            Alpha2LanguageCode.German,
+            Alpha2LanguageCode.French,
+            Alpha2LanguageCode.Dutch,
+            Alpha2LanguageCode.NorwegianBokmål,
+        ], "directionals.txt");
+        return this;
+    }
+    explore(span) {
+        if (span.flags.has("numeral"))
+            return;
+        if (this.index.has(span.normalized)) {
+            span.classifications.add("directional");
+            // try again for abbreviations denoted by a period such as 'n.'
+        }
+        else if (span.normalized.slice(-1) === "." && this.index.has(span.normalized.slice(0, -1))) {
+            span.classifications.add("directional");
+        }
+    }
+}
+//# sourceMappingURL=DirectionalClassifier.js.map

package/out/DirectionalClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"DirectionalClassifier.js","sourceRoot":"","sources":["../DirectionalClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,kBAAkB,EAAE,IAAI,EAAE,cAAc,EAAE,kBAAkB,EAAE,MAAM,iBAAiB,CAAA;AAE9F,MAAM,OAAO,qBAAsB,SAAQ,cAAc;IACjD,KAAK,CAAC,KAAK;QACjB,IAAI,CAAC,KAAK,GAAG,MAAM,kBAAkB,CACpC,IAAI,CAAC,SAAS,IAAI;YACjB,kBAAkB,CAAC,OAAO;YAC1B,kBAAkB,CAAC,OAAO;YAC1B,kBAAkB,CAAC,MAAM;YACzB,kBAAkB,CAAC,MAAM;YACzB,kBAAkB,CAAC,KAAK;YACxB,kBAAkB,CAAC,eAAe;SAClC,EACD,kBAAkB,CAClB,CAAA;QAED,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,OAAO,CAAC,IAAU;QACxB,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC;YAAE,OAAM;QAErC,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC;YACrC,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,aAAa,CAAC,CAAA;YAEvC,+DAA+D;QAChE,CAAC;aAAM,IAAI,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;YAC9F,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC,aAAa,CAAC,CAAA;QACxC,CAAC;IACF,CAAC;CACD"}

package/out/GivenNameClassifier.d.ts

@@ -0,0 +1,11 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { PhraseClassifier, Span } from "@mailwoman/core";
+export declare class GivenNameClassifier extends PhraseClassifier {
+    ready(): Promise<this>;
+    explore(span: Span): void;
+}
+//# sourceMappingURL=GivenNameClassifier.d.ts.map

package/out/GivenNameClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"GivenNameClassifier.d.ts","sourceRoot":"","sources":["../GivenNameClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,gBAAgB,EAAsB,IAAI,EAAkB,MAAM,iBAAiB,CAAA;AAE5F,qBAAa,mBAAoB,SAAQ,gBAAgB;IAC3C,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;IAc5B,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI;CAiBhC"}

package/out/GivenNameClassifier.js

@@ -0,0 +1,32 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { PhraseClassifier, prepareLocaleIndex, Span, TextNormalizer } from "@mailwoman/core";
+export class GivenNameClassifier extends PhraseClassifier {
+    async ready() {
+        this.index = await prepareLocaleIndex(this.languages ?? ["all"], "given_names.txt", new TextNormalizer({
+            lowercase: true,
+            // Omit short names.
+            minLength: 3,
+        }));
+        return this;
+    }
+    explore(span) {
+        if (span.flags.has("numeral"))
+            return;
+        const languages = this.index.get(span.normalized);
+        if (!languages)
+            return;
+        const firstChild = span.children.first || span;
+        const { previousSibling } = firstChild;
+        // if (previousSibling?.is("stop_word")) return
+        span.classifications.add({
+            classification: "given_name",
+            confidence: 1,
+            languages,
+        });
+    }
+}
+//# sourceMappingURL=GivenNameClassifier.js.map

package/out/GivenNameClassifier.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"GivenNameClassifier.js","sourceRoot":"","sources":["../GivenNameClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,gBAAgB,EAAE,kBAAkB,EAAE,IAAI,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAE5F,MAAM,OAAO,mBAAoB,SAAQ,gBAAgB;IACjD,KAAK,CAAC,KAAK;QACjB,IAAI,CAAC,KAAK,GAAG,MAAM,kBAAkB,CACpC,IAAI,CAAC,SAAS,IAAI,CAAC,KAAK,CAAC,EACzB,iBAAiB,EACjB,IAAI,cAAc,CAAC;YAClB,SAAS,EAAE,IAAI;YACf,oBAAoB;YACpB,SAAS,EAAE,CAAC;SACZ,CAAC,CACF,CAAA;QAED,OAAO,IAAI,CAAA;IACZ,CAAC;IAEM,OAAO,CAAC,IAAU;QACxB,IAAI,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC;YAAE,OAAM;QAErC,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;QACjD,IAAI,CAAC,SAAS;YAAE,OAAM;QAEtB,MAAM,UAAU,GAAG,IAAI,CAAC,QAAQ,CAAC,KAAK,IAAI,IAAI,CAAA;QAC9C,MAAM,EAAE,eAAe,EAAE,GAAG,UAAU,CAAA;QAEtC,+CAA+C;QAE/C,IAAI,CAAC,eAAe,CAAC,GAAG,CAAC;YACxB,cAAc,EAAE,YAAY;YAC5B,UAAU,EAAE,CAAC;YACb,SAAS;SACT,CAAC,CAAA;IACH,CAAC;CACD"}

package/out/HouseNumberClassifier.d.ts

@@ -0,0 +1,15 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { Span, WordClassifier } from "@mailwoman/core";
+/**
+ * Flags for house number classifications, used to distinguish between different types of house
+ * numbers.
+ */
+export type HouseNumberFlag = "alphanumeric" | "numeric" | "fractional" | "directional" | "separator" | "cyrillic";
+export declare class HouseNumberClassifier extends WordClassifier {
+    explore(span: Span): void;
+}
+//# sourceMappingURL=HouseNumberClassifier.d.ts.map

package/out/HouseNumberClassifier.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"HouseNumberClassifier.d.ts","sourceRoot":"","sources":["../HouseNumberClassifier.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,IAAI,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAmBtD;;;GAGG;AACH,MAAM,MAAM,eAAe,GAAG,cAAc,GAAG,SAAS,GAAG,YAAY,GAAG,aAAa,GAAG,WAAW,GAAG,UAAU,CAAA;AAoClH,qBAAa,qBAAsB,SAAQ,cAAc;IACjD,OAAO,CAAC,IAAI,EAAE,IAAI,GAAG,IAAI;CAoChC"}