@larsgw/formica 0.1.1 → 0.2.1

package/CHANGELOG.md

@@ -0,0 +1,53 @@
+## [0.2.1](https://github.com/identification-resources/formica/compare/v0.2.0...v0.2.1) (2023-08-10)
+
+
+### Bug Fixes
+
+* **resources:** do not validate old versions ([1a8dd2e](https://github.com/identification-resources/formica/commit/1a8dd2e0e373489e287d1fd89dbb7443245c5214))
+* **resources:** fix check for missing leaf taxa ([035f1ad](https://github.com/identification-resources/formica/commit/035f1ad20efa80b388ec3828d4ace4518c7b344e))
+
+
+
+# [0.2.0](https://github.com/identification-resources/formica/compare/v0.1.1...v0.2.0) (2023-08-09)
+
+
+### Features
+
+* **catalog:** add taxon_scope column ([6a06c37](https://github.com/identification-resources/formica/commit/6a06c37ec0640a8ffbc258ad6a4d81d84d35fee9))
+* **resources:** check for missing leaf taxa ([387c47d](https://github.com/identification-resources/formica/commit/387c47d10bb7f1fb09a68c10dfa869bb924b7633))
+
+
+
+## [0.1.1](https://github.com/identification-resources/formica/compare/v0.1.0...v0.1.1) (2023-05-10)
+
+
+### Bug Fixes
+
+* **processor:** add Node hashbang ([51dee57](https://github.com/identification-resources/formica/commit/51dee57b8547afba449b2b846bd7d65e7927368f))
+
+
+
+# [0.1.0](https://github.com/identification-resources/formica/compare/fed91fd6f350c47bd067d221a4d0e2278a199dae...v0.1.0) (2023-03-01)
+
+
+### Bug Fixes
+
+* **resources:** fix typo in taxon name pattern ([8f1b6cf](https://github.com/identification-resources/formica/commit/8f1b6cfa3858edb83c0aee589b84ad760ab815af))
+* **resources:** improve name parsing heuristics ([a40642e](https://github.com/identification-resources/formica/commit/a40642e9bdaeaae363584712e45f218809a5754b))
+
+
+### Features
+
+* **catalog:** add SDK for catalog data ([fed91fd](https://github.com/identification-resources/formica/commit/fed91fd6f350c47bd067d221a4d0e2278a199dae))
+* **csv:** allow custom delims in output ([b523614](https://github.com/identification-resources/formica/commit/b523614171b0ff96347cb876a08106df413c4032))
+* **processor:** processor for DwC creation ([0e6f21d](https://github.com/identification-resources/formica/commit/0e6f21dfe00a2892348a4ab04fcaf3d6c8324a91))
+* **processor:** processor for DwC indexing ([279773b](https://github.com/identification-resources/formica/commit/279773b085676cf7cb3eb4d81496ad219ac85bce))
+* **resources:** add SDK for resource text files ([0acd36d](https://github.com/identification-resources/formica/commit/0acd36dde04cc1240cb8be97d48812b201b2685a))
+* **resources:** add support for cluster markings ([112315c](https://github.com/identification-resources/formica/commit/112315c7ebd7151c4e2cc77ab20e0cec3f579db6))
+* **resources:** add support for diffs as input ([9751627](https://github.com/identification-resources/formica/commit/9751627a916f47552661596048f1f0d44d89b102))
+* **resources:** add support for hybrids ([e49e96a](https://github.com/identification-resources/formica/commit/e49e96af9c3958bab2a3b0508b1fc8a018b80716))
+* **validate:** add catalog validation script ([4cdf785](https://github.com/identification-resources/formica/commit/4cdf785afb2c77320cd9b5b76483e3464c0f1cff))
+* **validate:** add script to validate resources ([246fb5d](https://github.com/identification-resources/formica/commit/246fb5dd8da03e736884850874cc51a11be9985c))
+
+
+

package/lib/catalog/tables/work.js

@@ -44,6 +44,7 @@ var Work = /** @class */ (function (_super) {
             license: { required: false, multiple: true, format: value_1.FORMATS.LICENSE },
             key_type: { required: true, multiple: true, format: value_1.FORMATS.KEY_TYPE },
             taxon: { required: true, multiple: true },
+            taxon_scope: { required: false, multiple: true },
             scope: { required: false, multiple: true },
             region: { required: true, multiple: true },
             complete: { required: false, multiple: false, format: value_1.FORMATS.COMPLETE },

package/lib/resources/parse-text.js

@@ -43,6 +43,15 @@ var RANKS = [
     'race',
     'stirps' // not ICZN
 ];
+var MAIN_RANKS = [
+    'kingdom',
+    'phylum',
+    'class',
+    'order',
+    'family',
+    'genus',
+    'species'
+];
 var DWC_RANKS = [
     'kingdom',
     'phylum',
@@ -350,10 +359,7 @@ function parseHeader(header) {
     }
     return metadata;
 }
-function parseResource(resource) {
-    var _a = resource.split(/(\n---\n+)/), header = _a[0], _ = _a[1], rest = _a.slice(2);
-    var config = parseHeader(header);
-    var content = rest.join('');
+function validateResource(config, content) {
     // Check for too much indentation
     var longerIndent = new RegExp("^(  ){".concat(config.levels.length - 1, "}(?!  [+=>] ) "), 'm');
     var longerIndentMatch = content.match(longerIndent);
@@ -362,6 +368,23 @@ function parseResource(resource) {
         var line = (content.slice(0, offset).match(/\n/g) || []).length + 1;
         throw new SyntaxError("Too much indentation at ".concat(line, ":0\n").concat(content.slice(offset).split('\n', 1), "\n^"));
     }
+    // Check for missing leaf taxa
+    var leafTaxonRank = config.levels.filter(function (rank) { return MAIN_RANKS.includes(rank); }).pop();
+    var leafTaxonParentIndent = config.levels.indexOf(leafTaxonRank) - 1;
+    if (leafTaxonRank && leafTaxonParentIndent >= 0) {
+        var missingLeafTaxa = new RegExp("^((?:  ){0,".concat(leafTaxonParentIndent, "})(?![+=> ] ).*\\n(\\1(  )+[+=>].*\\n)*(?!\\1  )"), 'm');
+        var missingLeafTaxaMatch = content.match(missingLeafTaxa);
+        if (missingLeafTaxaMatch !== null) {
+            var offset = missingLeafTaxaMatch.index;
+            var line = (content.slice(0, offset).match(/\n/g) || []).length + 1;
+            throw new SyntaxError("Missing leaf taxon at ".concat(line, ":0\n").concat(content.slice(offset).split('\n', 1), "\n^"));
+        }
+    }
+}
+function parseResource(resource) {
+    var _a = resource.split(/(\n---\n+)/), header = _a[0], _ = _a[1], rest = _a.slice(2);
+    var config = parseHeader(header);
+    var content = rest.join('');
     return [config, content];
 }
 function parseResourceContent(content, resource, oldIds) {
@@ -468,6 +491,7 @@ function parseFile(file, id, old) {
     var oldResources = old ? splitResources(old.txt) : [];
     return splitResources(file).map(function (resource, index) {
         var _a = parseResource(resource), config = _a[0], content = _a[1];
+        validateResource(config, content);
         var template = {
             id: "".concat(id, ":").concat(index + 1),
             file: "".concat(id, "-").concat(index + 1),

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@larsgw/formica",
-  "version": "0.1.1",
+  "version": "0.2.1",
   "description": "SDK and tools for data from the Library of Identification Resources",
   "main": "lib/index.js",
   "types": "lib/index.d.ts",
@@ -11,9 +11,12 @@
     "loir-resources-index": "./lib/bin/process-resources-index.js"
   },
   "scripts": {
+    "test": "node --test --test-reporter spec",
     "lint": "eslint src",
     "build": "tsc -d",
+    "changelog": "conventional-changelog -p angular -i CHANGELOG.md -s -r 0",
     "preversion": "npm run lint",
+    "version": "npm run changelog && git add CHANGELOG.md",
     "prepublishOnly": "npm run build"
   },
   "repository": {
@@ -36,6 +39,7 @@
     "@types/node": "^18.14.1",
     "@typescript-eslint/eslint-plugin": "^5.54.0",
     "@typescript-eslint/parser": "^5.54.0",
+    "conventional-changelog-cli": "^3.0.0",
     "eslint": "^8.35.0",
     "typescript": "^4.9.5"
   }

package/src/catalog/tables/work.ts

@@ -26,6 +26,7 @@ export class Work extends Entity {
             license: { required: false, multiple: true, format: FORMATS.LICENSE },
             key_type: { required: true, multiple: true, format: FORMATS.KEY_TYPE },
             taxon: { required: true, multiple: true },
+            taxon_scope: { required: false, multiple: true },
             scope: { required: false, multiple: true },
             region: { required: true, multiple: true },
             complete: { required: false, multiple: false, format: FORMATS.COMPLETE },

package/src/resources/parse-text.ts

@@ -31,6 +31,16 @@ const RANKS: Rank[] = [
     'stirps' // not ICZN
 ]
 
+const MAIN_RANKS: Rank[] = [
+    'kingdom',
+    'phylum',
+    'class',
+    'order',
+    'family',
+    'genus',
+    'species'
+]
+
 const DWC_RANKS: DwcRank[] = [
     'kingdom',
     'phylum',
@@ -360,11 +370,7 @@ function parseHeader (header: string): ResourceMetadata {
     return metadata
 }
 
-function parseResource (resource: string): [ResourceMetadata, string] {
-    const [header, _, ...rest] = resource.split(/(\n---\n+)/)
-    const config = parseHeader(header)
-    const content = rest.join('')
-
+function validateResource (config: ResourceMetadata, content: string) {
     // Check for too much indentation
     const longerIndent = new RegExp(`^(  ){${config.levels.length - 1}}(?!  [+=>] ) `, 'm')
     const longerIndentMatch = content.match(longerIndent)
@@ -376,6 +382,27 @@ ${content.slice(offset).split('\n', 1)}
 ^`)
     }
 
+    // Check for missing leaf taxa
+    const leafTaxonRank = config.levels.filter(rank => MAIN_RANKS.includes(rank)).pop() as string
+    const leafTaxonParentIndent = config.levels.indexOf(leafTaxonRank) - 1
+    if (leafTaxonRank && leafTaxonParentIndent >= 0) {
+        const missingLeafTaxa = new RegExp(`^((?:  ){0,${leafTaxonParentIndent}})(?![+=> ] ).*\\n(\\1(  )+[+=>].*\\n)*(?!\\1  )`, 'm')
+        const missingLeafTaxaMatch = content.match(missingLeafTaxa)
+        if (missingLeafTaxaMatch !== null) {
+            const offset = missingLeafTaxaMatch.index
+            const line = (content.slice(0, offset).match(/\n/g) || []).length + 1
+            throw new SyntaxError(`Missing leaf taxon at ${line}:0
+${content.slice(offset).split('\n', 1)}
+^`)
+        }
+    }
+}
+
+function parseResource (resource: string): [ResourceMetadata, string] {
+    const [header, _, ...rest] = resource.split(/(\n---\n+)/)
+    const config = parseHeader(header)
+    const content = rest.join('')
+
     return [config, content]
 }
 
@@ -488,6 +515,7 @@ export function parseFile (file: string, id: WorkId, old?: ResourceHistory): Res
     const oldResources = old ? splitResources(old.txt) : []
     return splitResources(file).map((resource, index) => {
         const [config, content] = parseResource(resource)
+        validateResource(config, content)
         const template: Resource = {
             id: `${id}:${index + 1}`,
             file: `${id}-${index + 1}`,

package/test/resources.js

@@ -0,0 +1,19 @@
+const test = require('node:test')
+const assert = require('assert')
+
+const { resources } = require('../lib')
+
+test('resources', async (t) => {
+    await t.test('missing leaf taxa', (t) => {
+        assert.throws(() => {
+            resources.parseTextFile(`---
+levels: [family, genus, species]
+---
+
+Cydnidae
+Cydnidae
+  Legnotus
+    limbosus`, 'T1')
+        })
+    })
+})

package/lib/types.d.ts

@@ -1,62 +0,0 @@
-type Value = string[] | string;
-type SingleValue = string;
-interface FieldError {
-    field: string;
-    error: string;
-}
-interface WorkError extends FieldError {
-    entity: WorkId;
-}
-type Rank = string;
-type DwcRank = Rank;
-type TaxonStatus = string;
-type TaxonId = string;
-type ResourceId = string;
-type WorkId = string;
-interface WorkingTaxon {
-    scientificNameID?: TaxonId;
-    scientificName?: string;
-    scientificNameAuthorship?: string;
-    genericName?: string;
-    infragenericEpithet?: string;
-    specificEpithet?: string;
-    intraspecificEpithet?: string;
-    taxonRank?: Rank;
-    taxonRemarks?: string;
-    collectionCode?: ResourceId;
-    taxonomicStatus?: TaxonStatus;
-    acceptedNameUsageID?: TaxonId;
-    acceptedNameUsage?: string;
-    parentNameUsageID?: TaxonId;
-    parentNameUsage?: string;
-    kingdom?: string;
-    phylum?: string;
-    class?: string;
-    order?: string;
-    family?: string;
-    subfamily?: string;
-    genus?: string;
-    subgenus?: string;
-    higherClassification?: string;
-    scientificNameOnly?: string;
-    incorrect?: WorkingTaxon;
-}
-interface Taxon extends WorkingTaxon {
-    scientificNameID: TaxonId;
-    scientificName: string;
-    taxonRank: Rank;
-    collectionCode: ResourceId;
-    taxonomicStatus: string;
-}
-interface ResourceMetadata {
-    levels: Rank[];
-    scope: string[];
-    catalog?: object;
-}
-interface Resource {
-    id: string;
-    file: string;
-    workId: string;
-    metadata: ResourceMetadata;
-    taxa: Record<TaxonId, Taxon>;
-}