OpenSeaMap-dev Diskussion:Downloadable Raster Charts

ideas

web frontend

• die INT-chart Grenzen (als Basis fuer ein download-tool) bekomme ich inzwischen auf den Schirm: http://kap.grade.de/int-chart-ol2.html , sicherlich verbesserungsbeduerftig
• workflow
  • INT-Chart durch klick auf die Karte waehlen
  • download iniziieren
    • Weiterleitung auf neue Seite
    • pruefen, ob und in welchem Format vorhanden
    • wenn ja und nicht zu alt, downloadlinks anbieten
    • wenn nein Job in queue-db schreiben
      • number, name, bbox, scale, sessionid, ip-address, email, microtime, status=open, ...
• das ganze garnieren mit
  • Pruefung, ob das script fuer diese karte schon in der Warteschlange ist
  • Limitierung auf x downloads/jobs pro session ...
  • email notifier
  • captcha

prerequisites

• apache
• php
• sqlite

tile server

• cronjob checkt queue-db alle x Minuten
• waehlt aeltesten job aus queue-db mit status=open
• tiledownload ueber proxy mit landez
  • sinnvolles zoom-level anhand bbox errechnen (max. 10MB/chart)
  • mbtiles erzeugen (siehe auch mbtiles implementation)
  • stitchen und .png/.jpg erzeugen
  • .kap mit imgkap erstellen
  • ...
• status=done, zoom-level, date, ... in queue-db setzen
• temp-dir bereinigen und results in charts-dir kopieren

prerequisites

• apache
  • mod_python
• tilestache
• nginx remote proxy
• imagemagick
• python
  • landez
• imgkap

py script

• relies on landez

import logging
import shutil
import math
from landez import MBTilesBuilder
from landez import ImageExporter

logging.basicConfig(level=logging.DEBUG)

# to be imported from sqlite-queue
sqBottomGeoPre = 53.388333333333
sqTopGeoPre = 53.73
sqLeftGeoPre = 7.9483333333333
sqRightGeoPre = 8.35
sqNumTiles16 = 7656
sqChart = "1460"
sqname = "Die Jade, Innerer Teil"
sqUrl = "http://tiles.grade.de/tiles.py/openriverboatmap/{z}/{x}/{y}.png"
sqMicrotime= 1234

# to be calculated from numtiles16, numtiles < 700
# numtiles16 is right now hardcoded at int-chart.gejson for ol depth of view reasons, there might be smarter ways in future
# https://github.com/makinacorpus/landez/issues/44
myNumTiles = sqNumTiles16 * 4 * 4
myZoom = 18

while (myNumTiles > 10):
 myNumTiles = myNumTiles / 4
 myZoom = myZoom - 1

# settings
myFileName = sqChart
myMbtilesFile = myFileName + ".mbtiles"
myImageFile = myFileName + ".jpg"
myKapFile = myFileName + ".kap"
myTextFile = myFileName + ".txt"
myBbox = (sqLeftGeoPre, sqBottomGeoPre, sqRightGeoPre, sqTopGeoPre)
myZoomLevels = [myZoom]
myZoomLevel = myZoom
myTileSize = 256

# download and mbtiles build
mb = MBTilesBuilder(tiles_url=sqUrl, cache=False, filepath=myMbtilesFile, tile_size=myTileSize)
mb.add_coverage(myBbox, myZoomLevels)
mb.run(force=True)

# export image from mbtiles
# not the best idea: causes IO at temp dir, again
ie = ImageExporter(mbtiles_file=myMbtilesFile)
ie.export_image(myBbox, myZoomLevel, imagepath=myImageFile)
# save the geoinfo
myGrid = ie.grid_tiles(myBbox, myZoomLevel)
sqWidth = len(myGrid[0])
sqHeight = len(myGrid)
sqWidthPix = sqWidth * myTileSize
sqHeightPix = sqHeight * myTileSize

# delete temp dir (ImageExporter does not!)
# bad: path is hard coded. how do I get hold of landez's temp path?
# https://github.com/makinacorpus/landez/issues/42
try:
    shutil.rmtree('/tmp/landez')
except OSError:
    pass

# geoinfo output
logging.debug(" grid: " +  str(myGrid))

myLeftColumn = myGrid[0]
myTopLeft = myLeftColumn[0]
sqLeftTile = myTopLeft[0]
sqTopTile = myTopLeft[-1]

myRightColumn = myGrid[-1]
myBottomRight = myRightColumn[-1]
sqRightTile = myBottomRight[0]
sqBottomTile = myBottomRight[-1]

# calculate coords from tilenumbers
def num2deg(xtile, ytile, zoom):
  n = 2.0 ** zoom
  lon_deg = -180.0 + 360.0 * xtile / n
  lat_rad = math.atan(math.sinh(math.pi * (1.0 - 2.0 * (ytile / n))))
  lat_deg = 180.0 * lat_rad / math.pi
  return (lon_deg, lat_deg)

sqBottomGeoPost = (num2deg(sqRightTile + 1, sqBottomTile + 1, myZoom)[-1])
sqTopGeoPost = (num2deg(sqLeftTile, sqTopTile, myZoom)[-1])
sqLeftGeoPost = (num2deg(sqLeftTile, sqTopTile, myZoom)[0])
sqRightGeoPost = (num2deg(sqRightTile + 1, sqBottomTile + 1, myZoom)[0])

# shall be exported to sqlite-queue
logging.debug(" width: " + str(sqWidth) + " tiles")
logging.debug(" height: " + str(sqHeight) + " tiles")
logging.debug(" widthpix: " + str(sqWidthPix) + " pixel")
logging.debug(" heightpix: "+ str(sqHeightPix) + " pixel")

logging.debug(" left tile " + str(sqLeftTile))
logging.debug(" right tile: " + str(sqRightTile))
logging.debug(" top tile: " + str(sqTopTile))
logging.debug(" bottom tile: "+ str(sqBottomTile))

logging.debug(" left: " +  str(sqLeftGeoPost))
logging.debug(" right: " +  str(sqRightGeoPost))
logging.debug(" top: " +  str(sqTopGeoPost))
logging.debug(" bottom: " +  str(sqBottomGeoPost))

# start imgkap to generate .kap file
import subprocess
print "starting imgkap"
myImgkapCall = "./imgkap " + myImageFile + " " + str(sqTopGeoPost) + " " + str(sqLeftGeoPost) + " " + str(sqBottomGeoPost) + " " + str(sqRightGeoPost) + " " + myKapFile
print myImgkapCall
subprocess.call(myImgkapCall, shell=True)
print "end  imgkap"

Reason for ChartBundler

• Lets talk a bit about amounts here and not low stress or high stress. Using the charts offline on a mobile device. Lets say the display is about 1024x1024 Pixel, then we will need at least 16 tiles per map. If we start on zoom level 8 with the bounding box and continue to zoom level 16 for the harbours, we need around 2 Mio tiles. This 16 tiles on zoom level 8 is about the northern Adria. If we want the Channel just skipping Lands End up to the belgian coast, this would be 6 by 5 on zoom level 8. --Alexej

• Hi Alexej, do not understand this paragraph!? 2 test datasets of mine:
  • INT-1463 bei zoom-level 13 (s="53.7" n="54.25" w="7.4" e="9.05"):
    x: 37
    y: 21
    total: 777
    [1]
  • river Ruhr
    zoomlevel=16
    ~ 370 tiles
    [2]

--Kannix (Diskussion) 16:22, 29. Jan. 2015 (UTC)