@toolpath/tool-scraper 0.1.0 → 2.0.0

package/dist/vendors/regofix/scrape.js

@@ -53,11 +53,12 @@
  *    (`1715.08215` and `1725.08215`). This is why the nominal size is parsed
  *    from the vendor's own designation and `o_mm` is only ever a cross-check.
  */
-import { CAD_COLUMN, DIN_PREFIX } from '../../conventions.js';
+import { CAD_COLUMN, COLLET_SERIES_COLUMN, CONTACT_COLUMN, DIN_PREFIX, GAGE_COLUMNS, } from '../../conventions.js';
 import { VendorResponseError } from '../../errors.js';
 import { statusOf } from '../../fetch.js';
-import { consoleWarn } from '../../scrape.js';
-export const MM_PER_INCH = 25.4;
+import { convertLength, fractionValue } from '../../measure.js';
+import { compare } from '../../order.js';
+import { consoleWarn, unionHeader, } from '../../scrape.js';
 /**
  * The Searchkit proxy the ProductFinder posts its Elasticsearch queries to.
  * Discovered by reading the app bundle, which constructs
@@ -249,15 +250,17 @@ export function plain(value) {
         return '';
     return String(value);
 }
-/** `1/4` or `3.5` — the vendor prints both, and both are exact here. */
+/**
+ * `1/4` or `3.5` — the vendor prints both, and both are exact here.
+ *
+ * The grammar is `measure.fractionValue`'s; what stays here is the refusal.
+ * `3/` divides by zero and an absent regex capture arrives as `''`, and both
+ * would otherwise travel into a row as a nominal size — as would `0`, which
+ * reads as a number and is not a collet this vendor makes.
+ */
 export function parseSize(size) {
-    const value = size.includes('/')
-        ? Number(size.split('/')[0]) / Number(size.split('/')[1])
-        : Number(size);
-    // `3/` divides by zero and reaches here as Infinity, and an absent regex
-    // capture as `''` -> 0; both would travel into a row as a nominal size. Neither is a collet this
-    // vendor makes, so an unreadable size is refused rather than carried.
-    if (!Number.isFinite(value) || value <= 0) {
+    const value = fractionValue(size);
+    if (value === null || value <= 0) {
         throw new RangeError(`unrecognized size: ${JSON.stringify(size)}`);
     }
     return value;
@@ -336,9 +339,9 @@ export function holderRow(source, properties, warn = consoleWarn) {
     const row = {
         'Material Number': sku,
         'ISO Catalog Number': title,
-        CST: (parsed['series'] ?? '').replaceAll(' ', ''),
-        contact,
-        L1_mm: plain(gauge),
+        [COLLET_SERIES_COLUMN]: (parsed['series'] ?? '').replaceAll(' ', ''),
+        [CONTACT_COLUMN]: contact,
+        [GAGE_COLUMNS.millimeters]: plain(gauge),
         D2_mm: plain(pinned(properties, 'A1', sku)),
         B3_mm: plain(projection),
         [CAD_COLUMN]: cadUrl(source),
@@ -404,7 +407,7 @@ export function colletRow(source, warn = consoleWarn) {
     }
     const nominal = parseSize(parsed['size'] ?? '');
     const unit = inches ? 'inches' : 'millimeters';
-    const nominalMm = inches ? round(nominal * MM_PER_INCH, 6) : nominal;
+    const nominalMm = inches ? round(convertLength(nominal, 'inches', 'millimeters'), 6) : nominal;
     // A powRgrip collet clamps one size to h6 (h9 on the turning and tapping
     // lines) rather than closing over a range, so its capacity is its nominal
     // diameter at both ends. That is the vendor's `Clamping range or tolerance`
@@ -435,29 +438,11 @@ export function colletRow(source, warn = consoleWarn) {
         // Projected for the same reason `D1` is: an inch tapping collet whose
         // square sits only in `Square_in` is invisible to mm-side fit arithmetic.
         if (inches)
-            row['Square_mm'] = plain(round(square * MM_PER_INCH, 6));
+            row['Square_mm'] = plain(round(convertLength(square, 'inches', 'millimeters'), 6));
     }
     crossCheckOmm(row, nominalMm, warn);
     return row;
 }
-/**
- * Rows to a result whose header is the union of their keys, in first-seen
- * order.
- *
- * A union rather than the first row's keys: a mixed-unit collet family has
- * `D1_mm` on its metric rows and `D1_in` on its inch ones, and keying off row
- * one would drop whichever came second.
- */
-export function unionHeader(rows) {
-    const header = [];
-    for (const row of rows) {
-        for (const key of Object.keys(row)) {
-            if (!header.includes(key))
-                header.push(key);
-        }
-    }
-    return header;
-}
 /**
  * Every powRgrip holder of `group` whose taper is in {@link SCRAPED_TAPERS}.
  *
@@ -476,7 +461,7 @@ export async function scrapeHolders(fetcher, group = 'BT/PG', category = 'BT', o
     });
     const wanted = sources
         .filter((s) => SCRAPED_TAPERS.some((taper) => text(s, 'title').startsWith(taper)))
-        .sort((a, b) => text(a, 'field_sku_fulltext').localeCompare(text(b, 'field_sku_fulltext'), 'en'));
+        .sort((a, b) => compare(text(a, 'field_sku_fulltext'), text(b, 'field_sku_fulltext')));
     const rows = [];
     for (const source of wanted) {
         const sku = text(source, 'field_sku_fulltext');
@@ -506,7 +491,7 @@ export async function scrapeCollets(fetcher, group, sizes, options = {}) {
     });
     const wanted = sources
         .filter((s) => sizes.includes(String(one(s, 'norm_size'))))
-        .sort((a, b) => text(a, 'field_sku_fulltext').localeCompare(text(b, 'field_sku_fulltext'), 'en'));
+        .sort((a, b) => compare(text(a, 'field_sku_fulltext'), text(b, 'field_sku_fulltext')));
     return finish(
     // `options.warn` unguarded: `colletRow` owns the fallback, and defaulting
     // it here too would be two layers deciding the same thing.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@toolpath/tool-scraper",
-  "version": "0.1.0",
+  "version": "2.0.0",
   "description": "Scrape cutting-tool geometry from vendor catalogs into records and CSVs",
   "license": "MIT",
   "engines": {
@@ -43,6 +43,18 @@
       "types": "./dist/vendors/destinytool/index.d.ts",
       "import": "./dist/vendors/destinytool/index.js"
     },
+    "./vendors/emuge": {
+      "types": "./dist/vendors/emuge/index.d.ts",
+      "import": "./dist/vendors/emuge/index.js"
+    },
+    "./vendors/harvey": {
+      "types": "./dist/vendors/harvey/index.d.ts",
+      "import": "./dist/vendors/harvey/index.js"
+    },
+    "./vendors/maritool": {
+      "types": "./dist/vendors/maritool/index.d.ts",
+      "import": "./dist/vendors/maritool/index.js"
+    },
     "./registry": {
       "types": "./dist/registry.d.ts",
       "import": "./dist/registry.js"
@@ -50,6 +62,18 @@
     "./families": {
       "types": "./dist/families/index.d.ts",
       "import": "./dist/families/index.js"
+    },
+    "./families/harvey": {
+      "types": "./dist/families/harvey.d.ts",
+      "import": "./dist/families/harvey.js"
+    },
+    "./families/maritool": {
+      "types": "./dist/families/maritool.d.ts",
+      "import": "./dist/families/maritool.js"
+    },
+    "./families/emuge": {
+      "types": "./dist/families/emuge.d.ts",
+      "import": "./dist/families/emuge.js"
     }
   },
   "files": [
@@ -57,11 +81,6 @@
     "LICENSE",
     "README.md"
   ],
-  "scripts": {
-    "build": "tsc -p tsconfig.build.json",
-    "check-types": "tsc --noEmit",
-    "test": "vitest run"
-  },
   "dependencies": {
     "htmlparser2": "12.0.0"
   },
@@ -72,5 +91,10 @@
   },
   "bin": {
     "toolpath-scrape": "./dist/node/main.js"
+  },
+  "scripts": {
+    "build": "tsc -p tsconfig.build.json",
+    "check-types": "tsc --noEmit",
+    "test": "vitest run"
   }
-}
+}