@ca-plant-list/ca-plant-list 0.4.11 → 0.4.12

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@ca-plant-list/ca-plant-list",
-    "version": "0.4.11",
+    "version": "0.4.12",
     "description": "Tools to create Jekyll files for a website listing plants in an area of California.",
     "license": "MIT",
     "repository": {
@@ -16,7 +16,8 @@
     "bin": {
         "ca-plant-list": "scripts/build-site.js",
         "ca-plant-book": "scripts/build-ebook.js",
-        "cpl-photos": "scripts/cpl-photos.js"
+        "cpl-photos": "scripts/cpl-photos.js",
+        "cpl-tools": "scripts/cpl-tools.js"
     },
     "dependencies": {
         "archiver": "^5.3.1",
@@ -36,8 +37,8 @@
         "@types/markdown-it": "^14.1.2",
         "@types/node": "^22.7.8",
         "@types/unzipper": "^0.10.9",
-        "ajv-cli": "^5.0.0",
         "eslint": "^9.13.0",
+        "exceljs": "^4.4.0",
         "prettier": "^3.3.3",
         "typescript": "^5.6.3"
     }

package/schemas/exceptions.schema.json

@@ -9,12 +9,17 @@
                     "badjepsonid": {
                         "const": true
                     },
+                    "native": { "type": "boolean" },
                     "notintaxondata": {
                         "const": true
                     }
                 },
                 "additionalProperties": false
             },
+            "calscape": {
+                "type": "object",
+                "properties": { "notnative": { "const": true } }
+            },
             "comment": {
                 "type": "string"
             },
@@ -54,4 +59,4 @@
         },
         "additionalProperties": false
     }
-}
+}

package/scripts/cpl-photos.js

@@ -74,7 +74,11 @@ async function addMissingPhotos(options) {
         }
     }
 
-    CSV.writeFile(`${options.outputdir}/${PHOTO_FILE_NAME}`, data, headers);
+    CSV.writeFileArray(
+        `${options.outputdir}/${PHOTO_FILE_NAME}`,
+        data,
+        headers,
+    );
 }
 
 /**

package/scripts/cpl-tools.js

@@ -0,0 +1,177 @@
+#!/usr/bin/env node
+
+import * as path from "node:path";
+import { Option } from "commander";
+import { Taxa } from "../lib/taxa.js";
+import { Program } from "../lib/program.js";
+import { Calflora } from "../lib/tools/calflora.js";
+import { Exceptions } from "../lib/exceptions.js";
+import { ErrorLog } from "../lib/errorlog.js";
+import { Calscape } from "../lib/tools/calscape.js";
+
+const TOOLS = {
+    CALFLORA: "calflora",
+    CALSCAPE: "calscape",
+    INAT: "inat",
+    JEPSON_EFLORA: "jepson-eflora",
+    JEPSON_FAM: "jepson-families",
+    RPI: "rpi",
+    TEXT: "text",
+};
+
+const ALL_TOOLS = [
+    TOOLS.CALFLORA,
+    TOOLS.CALSCAPE,
+    TOOLS.INAT,
+    TOOLS.JEPSON_EFLORA,
+    TOOLS.RPI,
+    TOOLS.TEXT,
+];
+
+const OPT_LOADER = "loader";
+const OPT_TOOL = "tool";
+
+const TOOLS_DATA_DIR = "./external_data";
+
+/**
+ * @param {import("commander").Command} program
+ * @param {import("commander").OptionValues} options
+ */
+async function build(program, options) {
+    let tools = options[OPT_TOOL];
+    if (!tools) {
+        program.help();
+    }
+    if (tools[0] === "all") {
+        tools = ALL_TOOLS;
+    }
+
+    const exceptions = new Exceptions(options.datadir);
+    // const config = new Config(options.datadir);
+    const taxa = await getTaxa(options);
+
+    const errorLog = new ErrorLog(options.outputdir + "/log.tsv", true);
+    for (const tool of tools) {
+        switch (tool) {
+            case TOOLS.CALFLORA:
+                await Calflora.analyze(
+                    TOOLS_DATA_DIR,
+                    taxa,
+                    exceptions,
+                    errorLog,
+                );
+                break;
+            case TOOLS.CALSCAPE:
+                await Calscape.analyze(
+                    TOOLS_DATA_DIR,
+                    options.datadir,
+                    taxa,
+                    exceptions,
+                    errorLog,
+                    !!options.update,
+                );
+                break;
+            case TOOLS.INAT:
+                // await INat.analyze(
+                //     TOOLS_DATA_DIR,
+                //     taxa,
+                //     exceptions,
+                //     errorLog,
+                //     options.inTaxafile,
+                // );
+                break;
+            case TOOLS.JEPSON_EFLORA: {
+                // const eflora = new JepsonEFlora(
+                //     TOOLS_DATA_DIR,
+                //     taxa,
+                //     errorLog,
+                //     options.efLognotes,
+                // );
+                // await eflora.analyze(exceptions);
+                break;
+            }
+            case TOOLS.JEPSON_FAM:
+                // await JepsonFamilies.build(TOOLS_DATA_DIR, options.outputdir);
+                break;
+            case TOOLS.RPI:
+                // await RPI.analyze(
+                //     TOOLS_DATA_DIR,
+                //     taxa,
+                //     config,
+                //     exceptions,
+                //     errorLog,
+                // );
+                break;
+            case TOOLS.TEXT:
+                // SupplementalText.analyze(taxa, errorLog);
+                break;
+            default:
+                console.log("unrecognized tool: " + tool);
+                return;
+        }
+    }
+
+    errorLog.write();
+}
+
+/**
+ * @param {import("commander").OptionValues} options
+ */
+async function getTaxa(options) {
+    const errorLog = new ErrorLog(options.outputdir + "/errors.tsv", true);
+
+    const loader = options[OPT_LOADER];
+    let taxa;
+    if (loader) {
+        const taxaLoaderClass = await import("file:" + path.resolve(loader));
+        taxa = await taxaLoaderClass.TaxaLoader.loadTaxa(options, errorLog);
+    } else {
+        taxa = new Taxa(
+            Program.getIncludeList(options.datadir),
+            errorLog,
+            options.showFlowerErrors,
+        );
+    }
+
+    errorLog.write();
+    return taxa;
+}
+
+const program = Program.getProgram();
+program.addOption(
+    new Option(
+        "-t, --tool <tool...>",
+        "The tools to run. Value may be any subset of the tools below.",
+    ).choices(["all"].concat(ALL_TOOLS).concat(TOOLS.JEPSON_FAM)),
+);
+program.option(
+    "--in-taxafile <file>",
+    "The name of the file containing the iNaturalist taxa. Can be used for testing on a smaller subset of the iNaturalist data.",
+    "inat_taxa.csv",
+);
+program.option(
+    "--ef-lognotes",
+    "When running the jepson-eflora tool, include eFlora notes, invalid names, etc. in the log file.",
+);
+program.option(
+    "--loader <path>",
+    "The path (relative to the current directory) of the JavaScript file containing the TaxaLoader class. If not provided, the default TaxaLoader will be used.",
+);
+program.option("--update", "Update taxa.csv to remove errors if possible.");
+program.addHelpText(
+    "after",
+    `
+Tools:
+    'all' runs the 'calflora', 'inat', 'jepson-eflora', 'rpi', and 'text' tools.
+    '${TOOLS.CALFLORA}' retrieves data from Calflora and compares with local data.
+    '${TOOLS.CALSCAPE}' retrieves data from Calscape and compares with local data.
+    '${TOOLS.INAT}' retrieves data from iNaturalist and compares with local data.
+    '${TOOLS.JEPSON_EFLORA}' retrieves data from Jepson eFlora indexes and compares with local data.
+    '${TOOLS.JEPSON_FAM}' retrieves section, family and genus data from Jepson eFlora and creates data files for use by ca-plant-list.
+    '${TOOLS.RPI}' retrieves data from the CNPS Rare Plant Inventory and compares with local data.
+    '${TOOLS.TEXT}' checks supplemental text files to make sure their names are referenced.
+    `,
+);
+program.action((options) => build(program, options));
+
+await program.parseAsync();

package/types/classes.d.ts

@@ -83,6 +83,7 @@ declare class Taxa {
     getFlowerColors(): FlowerColor[];
     getTaxon(name: string): Taxon;
     getTaxonList(): Taxon[];
+    isSubset(): boolean;
 }
 
 declare class TaxaCol {
@@ -96,8 +97,11 @@ declare class Taxon {
     getBaseFileName(): string;
     getBloomEnd(): number | undefined;
     getBloomStart(): number | undefined;
+    getCalfloraID(): string;
     getCalfloraName(): string;
     getCalfloraTaxonLink(): string | undefined;
+    getCalscapeCommonName(): string | undefined;
+    getCalscapeName(): string;
     getCESA(): string | undefined;
     getCommonNames(): string[];
     getFamily(): Family;
@@ -123,6 +127,7 @@ declare class Taxon {
     getRPITaxonLink(): string;
     getStatusDescription(config: Config): string;
     getSynonyms(): string[];
+    isCANative(): boolean;
     isNative(): boolean;
 }
 
@@ -130,6 +135,7 @@ declare class TaxonData {
     bloom_end: string;
     bloom_start: string;
     calrecnum: string;
+    calscape_cn?: string;
     CESA: string;
     "common name": string;
     CRPR: string;