Merge unify_projection and CF cache changes to master

Merge Daniel's unify_projection change (with type fix) and
Solomon's clear cloudfront cache for COGs.

app/models/pydantic/creation_options.py

@@ -115,6 +115,13 @@ class FieldType(StrictBaseModel):
 
 
 class RasterTileSetAssetCreationOptions(StrictBaseModel):
+    unify_projection: bool = Field(
+        False,
+        description=(
+            "First re-project to a common projection (EPSG:4326). Necessary "
+            "when input files are in different projections from each other."
+        )
+    )
     pixel_meaning: str
     data_type: DataType
     nbits: Optional[int]

app/routes/datasets/versions.py

@@ -177,6 +177,7 @@ async def update_version(
                 AssetType.dynamic_vector_tile_cache,
                 AssetType.static_vector_tile_cache,
                 AssetType.raster_tile_cache,
+                AssetType.cog,
             ],
         )
 

app/tasks/raster_tile_set_assets/raster_tile_set_assets.py

@@ -18,9 +18,10 @@
 from app.models.pydantic.geostore import FeatureCollection
 from app.models.pydantic.jobs import Job
 from app.models.pydantic.statistics import BandStats, Histogram, RasterStats
+from app.settings.globals import DATA_LAKE_BUCKET
 from app.tasks import Callback, callback_constructor
 from app.tasks.batch import execute
-from app.tasks.raster_tile_set_assets.utils import create_pixetl_job
+from app.tasks.raster_tile_set_assets.utils import create_pixetl_job, create_unify_projection_job
 from app.utils.aws import get_s3_client
 from app.utils.path import (
     get_asset_uri,
@@ -67,13 +68,42 @@ async def raster_tile_set_asset(
 
     creation_options = PixETLCreationOptions(**co)
 
+    jobs: List[Job] = list()
     callback: Callback = callback_constructor(asset_id)
 
-    create_raster_tile_set_job: Job = await create_pixetl_job(
-        dataset, version, creation_options, "create_raster_tile_set", callback
+    unify_job: Optional[Job] = None
+    if creation_options.unify_projection:
+        target_crs = "epsg:4326"
+        new_src_uris = list()
+        for i,_ in enumerate(creation_options.source_uri):
+            new_src_uris.append(
+                f"s3://{DATA_LAKE_BUCKET}/{dataset}/{version}/raster/"
+                f"{target_crs.replace(':', '-')}/reprojected/SRC_{i}"
+            )
+        target_prefix = new_src_uris[0].rsplit("/", 1)[0]
+        unify_job = await create_unify_projection_job(
+            dataset,
+            creation_options.source_uri,
+            target_prefix,
+            target_crs,
+            "unify_projection",
+            callback
+        )
+        jobs.append(unify_job)
+        creation_options.source_uri = new_src_uris
+
+    jobs.append(
+        await create_pixetl_job(
+            dataset,
+            version,
+            creation_options,
+            "create_raster_tile_set",
+            callback,
+            [unify_job] if unify_job is not None else None,
+        )
     )
 
-    log: ChangeLog = await execute([create_raster_tile_set_job])
+    log: ChangeLog = await execute(jobs)
 
     return log
 

app/tasks/raster_tile_set_assets/utils.py

@@ -7,7 +7,7 @@
 from app.models.enum.assets import AssetType
 from app.models.enum.pixetl import ResamplingMethod
 from app.models.pydantic.creation_options import PixETLCreationOptions
-from app.models.pydantic.jobs import GDALDEMJob, Job, PixETLJob
+from app.models.pydantic.jobs import GDALDEMJob, Job, PixETLJob, GDAL2TilesJob
 from app.settings.globals import (
     AWS_GCS_KEY_SECRET_ARN,
     DEFAULT_JOB_DURATION,
@@ -225,3 +225,38 @@ async def create_resample_job(
         parents=[parent.job_name for parent in parents] if parents else None,
         **kwargs,
     )
+
+async def create_unify_projection_job(
+    dataset: str,
+    old_source_uris: List[str],
+    target_prefix: str,
+    target_crs: str,
+    job_name: str,
+    callback: Callback
+) -> GDAL2TilesJob:
+    """Creates a Batch job that takes all files indicated in old_source_uris
+    and re-projects each to a common CRS, then places them in a mirror of the
+    original directory structure under the target_prefix, divided by source
+    number. More specifically, the files from the first source URI will be
+    put at <target_prefix>/SRC_0, the files from the second under
+    <target_prefix>/SRC_1, and so on.
+    """
+
+    command = [
+        "unify_projection.sh",
+        "--target_crs",
+        target_crs,
+    ]
+
+    for s in old_source_uris:
+        command.extend(["--source", s])
+
+    command.extend(["--target", target_prefix])
+
+    return GDAL2TilesJob(
+        dataset=dataset,
+        job_name=job_name,
+        command=command,
+        environment=JOB_ENV,
+        callback=callback,
+    )

batch/gdal-python.dockerfile

@@ -1,4 +1,4 @@
-FROM globalforestwatch/data-api-gdal:v1.2.1
+FROM globalforestwatch/data-api-gdal:v1.2.2
 
 # Copy scripts
 COPY ./batch/scripts/ /opt/scripts/

batch/scripts/_tiff_crosses_dateline.sh

@@ -0,0 +1,40 @@
+#!/bin/bash
+#
+# USAGE: _tiff_crosses_dateline.sh raster_file
+#
+# Prints the string "true" if the input raster will cross the dateline
+# when converting to EPSG:4326, "false" otherwise
+#
+# Needs GDAL 2.0+ and Python
+#
+# Credit: Slightly modified from https://gis.stackexchange.com/a/222341
+
+
+if [ -z "${1}" ]; then
+    echo -e "Error: No input raster file given.\n> USAGE: _tiff_crosses_dateline.sh raster_file"
+    exit 1
+fi
+
+# Get raster info, save it to a variable as we need it several times
+gdalinfo=$(gdalinfo "${1}" -json)
+
+# Exit if -json switch is not available
+if [ ! -z $(echo $gdalinfo | grep "^Usage:") ]; then
+    echo -e "Error: GDAL command failed, Version 2.0+ is needed"
+    exit 1
+fi
+
+function jsonq {
+    echo "${1}" | python -c "import json,sys; jdata = sys.stdin.read(); data = json.loads(jdata); print(data${2});"
+}
+
+ulx=$(jsonq "$gdalinfo" "['wgs84Extent']['coordinates'][0][0][0]")
+llx=$(jsonq "$gdalinfo" "['wgs84Extent']['coordinates'][0][1][0]")
+lrx=$(jsonq "$gdalinfo" "['wgs84Extent']['coordinates'][0][3][0]")
+urx=$(jsonq "$gdalinfo" "['wgs84Extent']['coordinates'][0][2][0]")
+
+crossing_dateline=false
+test $(python -c "print(${ulx}>${lrx})") = True && crossing_dateline=true
+test $(python -c "print(${llx}>${urx})") = True && crossing_dateline=true
+
+echo -n "${crossing_dateline}"

batch/scripts/_warp_and_upload.sh

@@ -0,0 +1,38 @@
+#!/bin/bash
+
+set -e
+
+# arguments:
+# $0 - The name of this script
+# $1 - local_src_file
+# $2 - local_warped_file
+# $3 - target_crs
+# $4 - remote target file
+
+if aws s3 ls "$4"; then
+  echo "Remote target file $4 already exists, skipping..."
+  exit 0
+fi
+
+warp_options=("-co" "COMPRESS=DEFLATE" "-co" "TILED=yes")
+
+echo "Seeing if TIFF crosses the dateline"
+crosses="$(_tiff_crosses_dateline.sh $1)"
+if [ "${crosses}" = "true" ]; then
+  echo "$1 crosses the dateline"
+  warp_options+=("--config" "CENTER_LONG" "180")
+else
+  echo "$1 does not cross the dateline"
+fi
+
+echo "Now warping $1 to $2"
+gdalwarp "$1" "$2" -t_srs "$3" "${warp_options[@]}"
+echo "Done warping $1 to $2"
+
+echo "Now uploading $2 to $4"
+aws s3 cp --no-progress "$2" "$4"
+echo "Done uploading $2 to $4"
+
+echo "Finally, deleting local files $1 and $2"
+rm "$1" "$2"
+echo "Done deleting local files $1 and $2"

batch/scripts/get_arguments.sh

@@ -201,6 +201,11 @@ do
       shift # past argument
       shift # past value
       ;;
+      --target_crs)
+      TARGET_CRS="$2"
+      shift # past argument
+      shift # past value
+      ;;
       --target_bucket)
       TARGET_BUCKET="$2"
       shift # past argument

batch/scripts/resample.sh

@@ -6,7 +6,7 @@ set -e
 # -d | --dataset
 # -v | --version
 # -s | --source
-# -r | --resampling_method)
+# -r | --resampling_method
 # --zoom_level
 # -T | --target
 

batch/scripts/unify_projection.sh

@@ -0,0 +1,50 @@
+#!/bin/bash
+
+set -e
+
+# requires arguments
+# -s | --source
+# -T | --target
+#      --target_crs
+
+ME=$(basename "$0")
+. get_arguments.sh "$@"
+
+echo "Reproject to a common CRS"
+
+src_count=0
+CMD_ARGS=()
+
+for s in "${SRC[@]}"; do
+  source_dir="SRC_${src_count}"
+  mkdir -p "$source_dir"
+
+  echo "Now recursively downloading $s to $source_dir"
+  if [[ $s == gs://* ]]; then
+    gsutil -m cp -r "$s" "$source_dir"
+  elif [[ $s == s3://* ]]; then
+    aws s3 cp --recursive --no-progress "$s" "$source_dir"
+  fi
+  echo "Done downloading $s to $source_dir"
+
+  reprojected_dir="REPROJECTED_${src_count}"
+  mkdir -p "$reprojected_dir"
+
+  cd "${source_dir}"
+  for d in $(find . -type d | sed 's/.\///'); do
+    mkdir -p "../${reprojected_dir}/${d}"
+  done
+
+  for f in $(find . -iname "*.tif"| sed 's/.\///'); do
+    local_src_file="${source_dir}/${f}"
+    local_warped_file="${reprojected_dir}/${f}"
+    remote_target_file="${TARGET}/SRC_${src_count}/${f}"
+
+    CMD_ARGS+=("${local_src_file}" "${local_warped_file}" "${TARGET_CRS}" "${remote_target_file}")
+  done
+  cd ..
+
+  src_count=$(($src_count+1))
+done
+
+echo "${CMD_ARGS[@]}" | xargs -n 4 -P 32 _warp_and_upload.sh