Merge pull request #51 from Klimatbyran/feat/elasticsearch-explicit-m…

…appings

Feat/elasticsearch explicit mappings

src/elastic.ts

@@ -11,7 +11,7 @@ class Elastic {
     try  {
       this.client = new Client({ node });
       this.indexName = indexName;
-      this.pdfIndex = "pdfs";
+      this.pdfIndex = 'pdfs';
     } catch (error) {
       console.error('Elasticsearch constructor error:', error);
     }
@@ -58,7 +58,54 @@ class Elastic {
               properties: {
                 url: { type: 'keyword' },
                 pdfHash: { type: 'keyword' },
-                report: { type: 'nested' },
+                report: {
+                  type: 'nested',
+                  properties: {
+                    companyName: { type: 'keyword' },
+                    bransch: { type: 'keyword' },
+                    baseYear: { type: 'keyword' },
+                    url: { type: 'keyword' },
+                    emissions: {
+                      type: 'nested',
+                      properties: {
+                        year: { type: 'keyword' },
+                        scope1: {
+                          properties: {
+                            emissions: { type: 'double' },
+                            unit: { type: 'keyword' }
+                          }
+                        },
+                        scope2: {
+                          properties: {
+                            emissions: { type: 'double' },
+                            unit: { type: 'keyword' },
+                            mb: { type: 'double' },
+                            lb: { type: 'double' }
+                          }
+                        },
+                        scope3: {
+                          properties: {
+                            emissions: { type: 'double' },
+                            unit: { type: 'keyword' },
+                            categories: {
+                              type: 'nested',
+                              properties: {
+                                categoryName: { type: 'keyword' },
+                                emissions: { type: 'double' }
+                              }
+                            }
+                          }
+                        },
+                        totalEmissions: { type: 'double' },
+                        totalUnit: { type: 'keyword' },
+                      }
+                    },
+                    reliability: { type: 'keyword' },
+                    needsReview: { type: 'boolean' },
+                    reviewComment: { type: 'text' },
+                    reviewStatusCode: { type: 'keyword' }
+                  }
+                },
                 state: { type: 'keyword' },
                 timestamp: { type: 'date' }
               }

src/prompts/reflect.ts

@@ -7,6 +7,8 @@ Also convert the JSON to valid json and convert all units to metric ton CO2e. We
 
 **Data Output Format**: Present the extracted data in a structured JSON format. Include the year, Scope 1, Scope 2, Scope 3, and total emissions for each year.  If possible, also include the company's name and organization number in the JSON structure.
 
+**Important** Always generate this exact JSON structure, even if you cannot find the data. Indicate missing data with the error codes below, but make sure that the JSON structure is consistent. For example, if you cannot find the scope 3 categories you must make sure that the categories object is a valid JSON array.
+
     Example JSON structure:
 
     {