From 325efda28ee2238a89a4edf22031662515a732e0 Mon Sep 17 00:00:00 2001
From: jhpoelen <jhpoelen@xs4all.nl>
Date: Mon, 20 Jun 2022 16:45:57 -0500
Subject: [PATCH] add schema.jq file

---
 README.md |  6 ++----
 schema.jq | 27 +++++++++++++++++++++++++++
 2 files changed, 29 insertions(+), 4 deletions(-)
 create mode 100644 schema.jq

diff --git a/README.md b/README.md
index a2a51c5..53f9601 100644
--- a/README.md
+++ b/README.md
@@ -5,10 +5,8 @@ Machine readable version of Handbook of the Mammals of the World
 
 Physical copies of Handbook of the Mammals of the World are a valuable taxonomic resource for mammals.
 
-This repository uses Plazi digitized versions of these physical copies to generate [hmw.json](hmw.json),  and [hmw.csv](hmw.csv). For your convenience, the first 10 records can be found [hmw-sample.json](hmw-sample.json), [hmw-sample-pretty.json](hmw-sample-pretty.json) and [hmw-sample.csv](hmw-sample.csv). The json and csv files are derived from a Preston tracked version of Plazi's treatment bank available via https://github.com/plazi/treatments-xml. You can find a copy of a tracked copy of this in the [Releases](../releases) section of this repository.
-
-
-[Preston](https://preston.guoda.bio), [jq](https://stedolan.github.io/jq/), Miller (https://miller.readthedocs.io/en/latest/), and other commandline tools (e.g., ```time```, ```grep```, ```gzip```, ```tee```).
+This repository uses Plazi digitized versions of these physical copies to generate [hmw.json](hmw.json),  and [hmw.csv](hmw.csv). For your convenience, the first 10 records can be found [hmw-sample.json](hmw-sample.json), [hmw-sample-pretty.json](hmw-sample-pretty.json) and [hmw-sample.csv](hmw-sample.csv). The json and files are derived from a Preston tracked version of Plazi's treatment bank available via https://github.com/plazi/treatments-xml. You can find a copy of a tracked copy of this in the [Releases](../releases) section of this repository.
+The csv files are generated using  [jq](https://stedolan.github.io/jq/) using [schema.jq](schema.jq) and Miller (https://miller.readthedocs.io/en/latest/). 
 
 
 ```bash
diff --git a/schema.jq b/schema.jq
new file mode 100644
index 0000000..6367f07
--- /dev/null
+++ b/schema.jq
@@ -0,0 +1,27 @@
+{ 
+  "docId" : .docId,
+  "docOrigin" : .docOrigin,
+  "docISBN" : .docISBN,
+  "docName" : .docName,
+  "docMasterId" : .docMasterId,
+  "docPageNumber" : .docPageNumber,
+  "derivedFrom" : .["http://www.w3.org/ns/prov#wasDerivedFrom"],
+  "name" : .name,
+  "interpretedGenus" : .interpretedGenus,
+  "interpretedSpecies" : .interpretedSpecies,
+  "interpretedAuthorityName" : .interpretedAuthorityName,
+  "interpretedAuthorityYear" : .interpretedAuthorityYear,
+  "commonNames" : .commonNames,
+  "taxonomy" : .taxonomy,
+  "subspeciesAndDistribution" : .subspeciesAndDistribution,
+  "descriptiveNotes": .descriptiveNotes,
+  "habitat" : .habitat,
+  "foodAndFeeding" : .foodAndFeeding,
+  "breeding" : .breeding,
+  "activityPatterns" : .activityPatterns,
+  "movementsHomeRangeAndSocialOrganization" : .movementsHomeRangeAndSocialOrganization,
+  "statusAndConservation": .statusAndConservation,
+  "bibliography" : .bibliography,
+  "distributionImageURL" : .distributionImageURL,
+  "verbatimText" : .verbatimText
+}