Add validation for simple XPaths in form config

package.json

@@ -11,7 +11,7 @@
     "eslint": "eslint 'src/**/*.js' test/*.js 'test/**/*.js'",
     "docker-start-couchdb": "npm run docker-stop-couchdb && docker run -d -p 6984:5984 --rm --name cht-conf-couchdb couchdb:2.3.1 && sh test/scripts/wait_for_response_code.sh 6984 200 CouchDB",
     "docker-stop-couchdb": "docker stop cht-conf-couchdb || true",
-    "test": "npm run eslint && npm run docker-start-couchdb && npm run clean && mkdir -p build/test && cp -r test/data build/test/data && cd build/test && nyc --reporter=html mocha --forbid-only ../../test/**/*.spec.js && cd ../.. && npm run docker-stop-couchdb",
+    "test": "npm run eslint && npm run docker-start-couchdb && npm run clean && mkdir -p build/test && cp -r test/data build/test/data && cd build/test && nyc --reporter=html mocha --forbid-only \"../../test/**/*.spec.js\" && cd ../.. && npm run docker-stop-couchdb",
     "semantic-release": "semantic-release"
   },
   "bin": {
@@ -38,6 +38,7 @@
     "@hapi/joi": "^16.1.8",
     "@medic/translation-checker": "^1.0.1",
     "@parcel/watcher": "^2.0.5",
+    "@xmldom/xmldom": "^0.8.2",
     "canonical-json": "0.0.4",
     "csv-parse": "^4.16.0",
     "dom-compare": "^0.6.0",
@@ -67,7 +68,7 @@
     "terser-webpack-plugin": "^1.4.3",
     "uuid": "^8.3.2",
     "webpack": "^4.46.0",
-    "xmldom": "^0.6.0"
+    "xpath": "0.0.32"
   },
   "devDependencies": {
     "@commitlint/cli": "^13.2.0",

src/lib/forms-utils.js

@@ -1,5 +1,14 @@
+const xpath = require('xpath');
 const fs = require('./sync-fs');
 
+const XPATH_MODEL = '/h:html/h:head/model';
+
+const getNode = (currentNode, path) =>
+  xpath.parse(path).select1({ node: currentNode, allowAnyNamespaceForNoPrefix: true });
+
+const getNodes = (currentNode, path) =>
+  xpath.parse(path).select({ node: currentNode, allowAnyNamespaceForNoPrefix: true });
+
 module.exports = {
   /**
    * Get the full path of the form, or null if the path doesn't exist.
@@ -29,32 +38,61 @@ module.exports = {
     };
   },
 
-  // This isn't really how to parse XML, but we have fairly good control over the
-  // input and this code is working so far.  This may break with changes to the
-  // formatting of output from xls2xform.
+  /**
+   * Returns the node from the form XML specified by the given XPath.
+   * @param {Element} currentNode the current node in the form XML document
+   * @param {string} path the XPath expression
+   * @returns {Element} the selected node or `undefined` if not found
+   */
+  getNode,
+
+  /**
+   * Returns the nodes from the form XML specified by the given XPath.
+   * @param {Element} currentNode the current node in the form XML document
+   * @param {string} path the XPath expression
+   * @returns {Element} the selected nodes or an empty array if none are found
+   */
+  getNodes,
+
+  /**
+   * Returns the `bind` nodes for the given form XML.
+   * @param {Document} xmlDoc the form XML document
+   * @returns {Element}
+   */
+  getBindNodes: xmlDoc => getNodes(xmlDoc, `${XPATH_MODEL}/bind`),
+
+  /**
+   * Returns the primary (first) `instance` node for the given form XML.
+   * @param {Document} xmlDoc the form XML document
+   * @returns {Element}
+   */
+  getPrimaryInstanceNode: xmlDoc => getNode(xmlDoc, `${XPATH_MODEL}/instance`),
 
   /**
    * Check whether the XForm has the <instanceID/> tag.
-   * @param {string} xml the XML string
+   * @param {string} xmlDoc the XML document
    * @returns {boolean}
    */
-  formHasInstanceId: xml => xml.includes('<instanceID/>'),
+  formHasInstanceId: xmlDoc => getNode(xmlDoc, `//meta/instanceID`) !== undefined,
 
+  // This isn't really how to parse XML, but we have fairly good control over the
+  // input and this code is working so far.  This may break with changes to the
+  // formatting of output from xls2xform.
   /**
    * Get the title string inside the <h:title> tag
    * @param {string} xml the XML string
    * @returns {string}
    */
   readTitleFrom: xml =>
-      xml.substring(xml.indexOf('<h:title>') + 9, xml.indexOf('</h:title>')),
+    xml.substring(xml.indexOf('<h:title>') + 9, xml.indexOf('</h:title>')),
 
   /**
    * Get the ID of the form
    * @param {string} xml the XML string
    * @returns {string}
    */
   readIdFrom: xml =>
-      xml.match(/<model>[^]*<\/model>/)[0]
-        .match(/<instance>[^]*<\/instance>/)[0]
-        .match(/id="([^"]*)"/)[1],
+    xml.match(/<model>[^]*<\/model>/)[0]
+      .match(/<instance>[^]*<\/instance>/)[0]
+      .match(/id="([^"]*)"/)[1],
 };

src/lib/validate-forms.js

@@ -1,3 +1,4 @@
+const { DOMParser } = require('@xmldom/xmldom');
 const argsFormFilter = require('./args-form-filter');
 const environment = require('./environment');
 const fs = require('./sync-fs');
@@ -7,6 +8,7 @@ const {
   getFormFilePaths
 } = require('./forms-utils');
 
+const domParser = new DOMParser();
 const VALIDATIONS_PATH = fs.path.resolve(__dirname, './validation/form');
 const validations = fs.readdir(VALIDATIONS_PATH)
   .filter(name => name.endsWith('.js'))
@@ -52,7 +54,7 @@ module.exports = async (projectDir, subDirectory, options={}) => {
     const { xformPath } = getFormFilePaths(formsDir, fileName);
     const xml = fs.read(xformPath);
 
-    const valParams = { xformPath, xmlStr: xml };
+    const valParams = { xformPath, xmlStr: xml, xmlDoc: domParser.parseFromString(xml) };
     for(const validation of validations) {
       if(validation.requiresInstance && !instanceProvided) {
         validationSkipped = true;